Upload folder using huggingface_hub
Browse files- checkpoints/checkpoint-10000/config.json +28 -0
- checkpoints/checkpoint-10000/model.safetensors +3 -0
- checkpoints/checkpoint-10000/optimizer.pt +3 -0
- checkpoints/checkpoint-10000/rng_state.pth +3 -0
- checkpoints/checkpoint-10000/scheduler.pt +3 -0
- checkpoints/checkpoint-10000/trainer_state.json +1421 -0
- checkpoints/checkpoint-10000/training_args.bin +3 -0
- checkpoints/checkpoint-11000/config.json +28 -0
- checkpoints/checkpoint-11000/model.safetensors +3 -0
- checkpoints/checkpoint-11000/optimizer.pt +3 -0
- checkpoints/checkpoint-11000/rng_state.pth +3 -0
- checkpoints/checkpoint-11000/scheduler.pt +3 -0
- checkpoints/checkpoint-11000/trainer_state.json +1561 -0
- checkpoints/checkpoint-11000/training_args.bin +3 -0
- checkpoints/checkpoint-12000/config.json +28 -0
- checkpoints/checkpoint-12000/model.safetensors +3 -0
- checkpoints/checkpoint-12000/optimizer.pt +3 -0
- checkpoints/checkpoint-12000/rng_state.pth +3 -0
- checkpoints/checkpoint-12000/scheduler.pt +3 -0
- checkpoints/checkpoint-12000/trainer_state.json +1701 -0
- checkpoints/checkpoint-12000/training_args.bin +3 -0
- checkpoints/checkpoint-13000/config.json +28 -0
- checkpoints/checkpoint-13000/model.safetensors +3 -0
- checkpoints/checkpoint-13000/optimizer.pt +3 -0
- checkpoints/checkpoint-13000/rng_state.pth +3 -0
- checkpoints/checkpoint-13000/scheduler.pt +3 -0
- checkpoints/checkpoint-13000/trainer_state.json +1841 -0
- checkpoints/checkpoint-13000/training_args.bin +3 -0
- checkpoints/checkpoint-14000/config.json +28 -0
- checkpoints/checkpoint-14000/model.safetensors +3 -0
- checkpoints/checkpoint-14000/optimizer.pt +3 -0
- checkpoints/checkpoint-14000/rng_state.pth +3 -0
- checkpoints/checkpoint-14000/scheduler.pt +3 -0
- checkpoints/checkpoint-14000/trainer_state.json +1981 -0
- checkpoints/checkpoint-14000/training_args.bin +3 -0
- checkpoints/config.json +28 -0
- checkpoints/model.safetensors +3 -0
- checkpoints/training_args.bin +3 -0
checkpoints/checkpoint-10000/config.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"AlbertForMaskedLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0,
|
| 6 |
+
"bos_token_id": 2,
|
| 7 |
+
"classifier_dropout_prob": 0.1,
|
| 8 |
+
"embedding_size": 128,
|
| 9 |
+
"eos_token_id": 3,
|
| 10 |
+
"hidden_act": "gelu_new",
|
| 11 |
+
"hidden_dropout_prob": 0,
|
| 12 |
+
"hidden_size": 768,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"inner_group_num": 1,
|
| 15 |
+
"intermediate_size": 3072,
|
| 16 |
+
"layer_norm_eps": 1e-12,
|
| 17 |
+
"max_position_embeddings": 512,
|
| 18 |
+
"model_type": "albert",
|
| 19 |
+
"num_attention_heads": 12,
|
| 20 |
+
"num_hidden_groups": 1,
|
| 21 |
+
"num_hidden_layers": 12,
|
| 22 |
+
"pad_token_id": 0,
|
| 23 |
+
"position_embedding_type": "absolute",
|
| 24 |
+
"torch_dtype": "float32",
|
| 25 |
+
"transformers_version": "4.38.2",
|
| 26 |
+
"type_vocab_size": 2,
|
| 27 |
+
"vocab_size": 30000
|
| 28 |
+
}
|
checkpoints/checkpoint-10000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4935ca50e26f57ed0c3511d14103e268f6b89205ef598adb0af7218b97ff4929
|
| 3 |
+
size 44890256
|
checkpoints/checkpoint-10000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:30148c53b8836ab81f89468f501ea830c4e0e55e5bd787c0511f1eaadfdd980f
|
| 3 |
+
size 89797322
|
checkpoints/checkpoint-10000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:edcbcc26743f028193a9edf550c3acd511f3b00b86c980be936032339e7e110c
|
| 3 |
+
size 14244
|
checkpoints/checkpoint-10000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2494212a7fdf911fca46666b1f4d72fd5601c979dd38224f107f835a89dd7be5
|
| 3 |
+
size 1064
|
checkpoints/checkpoint-10000/trainer_state.json
ADDED
|
@@ -0,0 +1,1421 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.11088564363571848,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 10000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"grad_norm": 3.2580924034118652,
|
| 14 |
+
"learning_rate": 2.2172949002217296e-06,
|
| 15 |
+
"loss": 10.2933,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.0,
|
| 20 |
+
"grad_norm": 2.4347386360168457,
|
| 21 |
+
"learning_rate": 4.434589800443459e-06,
|
| 22 |
+
"loss": 10.1894,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.0,
|
| 27 |
+
"grad_norm": 2.3895885944366455,
|
| 28 |
+
"learning_rate": 6.651884700665188e-06,
|
| 29 |
+
"loss": 10.1424,
|
| 30 |
+
"step": 150
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.0,
|
| 34 |
+
"grad_norm": 2.129647731781006,
|
| 35 |
+
"learning_rate": 8.869179600886918e-06,
|
| 36 |
+
"loss": 10.0995,
|
| 37 |
+
"step": 200
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.0,
|
| 41 |
+
"grad_norm": 2.3564186096191406,
|
| 42 |
+
"learning_rate": 1.1086474501108649e-05,
|
| 43 |
+
"loss": 10.0479,
|
| 44 |
+
"step": 250
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.0,
|
| 48 |
+
"grad_norm": 1.830551028251648,
|
| 49 |
+
"learning_rate": 1.3303769401330377e-05,
|
| 50 |
+
"loss": 9.9971,
|
| 51 |
+
"step": 300
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.0,
|
| 55 |
+
"grad_norm": 2.1173911094665527,
|
| 56 |
+
"learning_rate": 1.5521064301552106e-05,
|
| 57 |
+
"loss": 9.9201,
|
| 58 |
+
"step": 350
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.0,
|
| 62 |
+
"grad_norm": 1.6636557579040527,
|
| 63 |
+
"learning_rate": 1.7738359201773837e-05,
|
| 64 |
+
"loss": 9.8562,
|
| 65 |
+
"step": 400
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.0,
|
| 69 |
+
"grad_norm": 2.4503839015960693,
|
| 70 |
+
"learning_rate": 1.9955654101995567e-05,
|
| 71 |
+
"loss": 9.7599,
|
| 72 |
+
"step": 450
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.01,
|
| 76 |
+
"grad_norm": 1.822424054145813,
|
| 77 |
+
"learning_rate": 2.2172949002217298e-05,
|
| 78 |
+
"loss": 9.6608,
|
| 79 |
+
"step": 500
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.01,
|
| 83 |
+
"grad_norm": 1.6598998308181763,
|
| 84 |
+
"learning_rate": 2.4390243902439026e-05,
|
| 85 |
+
"loss": 9.55,
|
| 86 |
+
"step": 550
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.01,
|
| 90 |
+
"grad_norm": 1.8471707105636597,
|
| 91 |
+
"learning_rate": 2.6607538802660753e-05,
|
| 92 |
+
"loss": 9.4606,
|
| 93 |
+
"step": 600
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.01,
|
| 97 |
+
"grad_norm": 1.4833533763885498,
|
| 98 |
+
"learning_rate": 2.8824833702882487e-05,
|
| 99 |
+
"loss": 9.3283,
|
| 100 |
+
"step": 650
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.01,
|
| 104 |
+
"grad_norm": 1.688541054725647,
|
| 105 |
+
"learning_rate": 3.104212860310421e-05,
|
| 106 |
+
"loss": 9.2229,
|
| 107 |
+
"step": 700
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.01,
|
| 111 |
+
"grad_norm": 1.6466543674468994,
|
| 112 |
+
"learning_rate": 3.325942350332594e-05,
|
| 113 |
+
"loss": 9.1093,
|
| 114 |
+
"step": 750
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.01,
|
| 118 |
+
"grad_norm": 1.4169293642044067,
|
| 119 |
+
"learning_rate": 3.547671840354767e-05,
|
| 120 |
+
"loss": 8.9703,
|
| 121 |
+
"step": 800
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.01,
|
| 125 |
+
"grad_norm": 1.7079193592071533,
|
| 126 |
+
"learning_rate": 3.7694013303769404e-05,
|
| 127 |
+
"loss": 8.8351,
|
| 128 |
+
"step": 850
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.01,
|
| 132 |
+
"grad_norm": 1.5513204336166382,
|
| 133 |
+
"learning_rate": 3.9911308203991135e-05,
|
| 134 |
+
"loss": 8.7111,
|
| 135 |
+
"step": 900
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.01,
|
| 139 |
+
"grad_norm": 1.485573172569275,
|
| 140 |
+
"learning_rate": 4.212860310421286e-05,
|
| 141 |
+
"loss": 8.5627,
|
| 142 |
+
"step": 950
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.01,
|
| 146 |
+
"grad_norm": 1.511690616607666,
|
| 147 |
+
"learning_rate": 4.4345898004434597e-05,
|
| 148 |
+
"loss": 8.5042,
|
| 149 |
+
"step": 1000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.01,
|
| 153 |
+
"grad_norm": 2.1478614807128906,
|
| 154 |
+
"learning_rate": 4.656319290465632e-05,
|
| 155 |
+
"loss": 8.3287,
|
| 156 |
+
"step": 1050
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.01,
|
| 160 |
+
"grad_norm": 1.4060652256011963,
|
| 161 |
+
"learning_rate": 4.878048780487805e-05,
|
| 162 |
+
"loss": 8.2341,
|
| 163 |
+
"step": 1100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.01,
|
| 167 |
+
"grad_norm": 1.3950035572052002,
|
| 168 |
+
"learning_rate": 5.099778270509978e-05,
|
| 169 |
+
"loss": 8.1277,
|
| 170 |
+
"step": 1150
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.01,
|
| 174 |
+
"grad_norm": 1.5197688341140747,
|
| 175 |
+
"learning_rate": 5.3215077605321506e-05,
|
| 176 |
+
"loss": 8.0311,
|
| 177 |
+
"step": 1200
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.01,
|
| 181 |
+
"grad_norm": 1.3406693935394287,
|
| 182 |
+
"learning_rate": 5.543237250554324e-05,
|
| 183 |
+
"loss": 7.9824,
|
| 184 |
+
"step": 1250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.01,
|
| 188 |
+
"grad_norm": 1.4520119428634644,
|
| 189 |
+
"learning_rate": 5.7649667405764975e-05,
|
| 190 |
+
"loss": 7.9948,
|
| 191 |
+
"step": 1300
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.01,
|
| 195 |
+
"grad_norm": 1.179124116897583,
|
| 196 |
+
"learning_rate": 5.98669623059867e-05,
|
| 197 |
+
"loss": 7.9144,
|
| 198 |
+
"step": 1350
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.02,
|
| 202 |
+
"grad_norm": 1.4039533138275146,
|
| 203 |
+
"learning_rate": 6.208425720620842e-05,
|
| 204 |
+
"loss": 7.8768,
|
| 205 |
+
"step": 1400
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.02,
|
| 209 |
+
"grad_norm": 1.5542700290679932,
|
| 210 |
+
"learning_rate": 6.430155210643016e-05,
|
| 211 |
+
"loss": 7.894,
|
| 212 |
+
"step": 1450
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.02,
|
| 216 |
+
"grad_norm": 1.4150550365447998,
|
| 217 |
+
"learning_rate": 6.651884700665188e-05,
|
| 218 |
+
"loss": 7.8409,
|
| 219 |
+
"step": 1500
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.02,
|
| 223 |
+
"grad_norm": 1.6647827625274658,
|
| 224 |
+
"learning_rate": 6.873614190687362e-05,
|
| 225 |
+
"loss": 7.91,
|
| 226 |
+
"step": 1550
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.02,
|
| 230 |
+
"grad_norm": 1.7795697450637817,
|
| 231 |
+
"learning_rate": 7.095343680709535e-05,
|
| 232 |
+
"loss": 7.8256,
|
| 233 |
+
"step": 1600
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.02,
|
| 237 |
+
"grad_norm": 1.933110237121582,
|
| 238 |
+
"learning_rate": 7.317073170731707e-05,
|
| 239 |
+
"loss": 7.8463,
|
| 240 |
+
"step": 1650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.02,
|
| 244 |
+
"grad_norm": 1.1942570209503174,
|
| 245 |
+
"learning_rate": 7.538802660753881e-05,
|
| 246 |
+
"loss": 7.7827,
|
| 247 |
+
"step": 1700
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.02,
|
| 251 |
+
"grad_norm": 1.6759297847747803,
|
| 252 |
+
"learning_rate": 7.760532150776053e-05,
|
| 253 |
+
"loss": 7.8,
|
| 254 |
+
"step": 1750
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.02,
|
| 258 |
+
"grad_norm": 1.093256950378418,
|
| 259 |
+
"learning_rate": 7.982261640798227e-05,
|
| 260 |
+
"loss": 7.7461,
|
| 261 |
+
"step": 1800
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.02,
|
| 265 |
+
"grad_norm": 1.567872166633606,
|
| 266 |
+
"learning_rate": 8.2039911308204e-05,
|
| 267 |
+
"loss": 7.7338,
|
| 268 |
+
"step": 1850
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.02,
|
| 272 |
+
"grad_norm": 1.3017679452896118,
|
| 273 |
+
"learning_rate": 8.425720620842572e-05,
|
| 274 |
+
"loss": 7.804,
|
| 275 |
+
"step": 1900
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.02,
|
| 279 |
+
"grad_norm": 1.7510960102081299,
|
| 280 |
+
"learning_rate": 8.647450110864746e-05,
|
| 281 |
+
"loss": 7.7405,
|
| 282 |
+
"step": 1950
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.02,
|
| 286 |
+
"grad_norm": 1.7215120792388916,
|
| 287 |
+
"learning_rate": 8.869179600886919e-05,
|
| 288 |
+
"loss": 7.7429,
|
| 289 |
+
"step": 2000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.02,
|
| 293 |
+
"grad_norm": 1.6202715635299683,
|
| 294 |
+
"learning_rate": 9.090909090909092e-05,
|
| 295 |
+
"loss": 7.6588,
|
| 296 |
+
"step": 2050
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.02,
|
| 300 |
+
"grad_norm": 1.5680756568908691,
|
| 301 |
+
"learning_rate": 9.312638580931264e-05,
|
| 302 |
+
"loss": 7.6224,
|
| 303 |
+
"step": 2100
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.02,
|
| 307 |
+
"grad_norm": 1.462240219116211,
|
| 308 |
+
"learning_rate": 9.534368070953438e-05,
|
| 309 |
+
"loss": 7.6851,
|
| 310 |
+
"step": 2150
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.02,
|
| 314 |
+
"grad_norm": 2.2018320560455322,
|
| 315 |
+
"learning_rate": 9.75609756097561e-05,
|
| 316 |
+
"loss": 7.6443,
|
| 317 |
+
"step": 2200
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.02,
|
| 321 |
+
"grad_norm": 1.9520208835601807,
|
| 322 |
+
"learning_rate": 9.977827050997783e-05,
|
| 323 |
+
"loss": 7.6456,
|
| 324 |
+
"step": 2250
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.03,
|
| 328 |
+
"grad_norm": 1.115421175956726,
|
| 329 |
+
"learning_rate": 0.00010199556541019956,
|
| 330 |
+
"loss": 7.5894,
|
| 331 |
+
"step": 2300
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.03,
|
| 335 |
+
"grad_norm": 1.6002250909805298,
|
| 336 |
+
"learning_rate": 0.0001042128603104213,
|
| 337 |
+
"loss": 7.6017,
|
| 338 |
+
"step": 2350
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.03,
|
| 342 |
+
"grad_norm": 1.6516796350479126,
|
| 343 |
+
"learning_rate": 0.00010643015521064301,
|
| 344 |
+
"loss": 7.4548,
|
| 345 |
+
"step": 2400
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.03,
|
| 349 |
+
"grad_norm": 2.2168257236480713,
|
| 350 |
+
"learning_rate": 0.00010864745011086475,
|
| 351 |
+
"loss": 7.5867,
|
| 352 |
+
"step": 2450
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.03,
|
| 356 |
+
"grad_norm": 1.5447593927383423,
|
| 357 |
+
"learning_rate": 0.00011086474501108647,
|
| 358 |
+
"loss": 7.5317,
|
| 359 |
+
"step": 2500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.03,
|
| 363 |
+
"grad_norm": 1.6840906143188477,
|
| 364 |
+
"learning_rate": 0.00011308203991130821,
|
| 365 |
+
"loss": 7.5127,
|
| 366 |
+
"step": 2550
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.03,
|
| 370 |
+
"grad_norm": 1.2965503931045532,
|
| 371 |
+
"learning_rate": 0.00011529933481152995,
|
| 372 |
+
"loss": 7.4911,
|
| 373 |
+
"step": 2600
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.03,
|
| 377 |
+
"grad_norm": 1.643584966659546,
|
| 378 |
+
"learning_rate": 0.00011751662971175166,
|
| 379 |
+
"loss": 7.4416,
|
| 380 |
+
"step": 2650
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.03,
|
| 384 |
+
"grad_norm": 1.5419111251831055,
|
| 385 |
+
"learning_rate": 0.0001197339246119734,
|
| 386 |
+
"loss": 7.4944,
|
| 387 |
+
"step": 2700
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.03,
|
| 391 |
+
"grad_norm": 1.7774205207824707,
|
| 392 |
+
"learning_rate": 0.00012195121951219512,
|
| 393 |
+
"loss": 7.4244,
|
| 394 |
+
"step": 2750
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.03,
|
| 398 |
+
"grad_norm": 2.1709322929382324,
|
| 399 |
+
"learning_rate": 0.00012416851441241685,
|
| 400 |
+
"loss": 7.371,
|
| 401 |
+
"step": 2800
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.03,
|
| 405 |
+
"grad_norm": 1.5503411293029785,
|
| 406 |
+
"learning_rate": 0.0001263858093126386,
|
| 407 |
+
"loss": 7.3031,
|
| 408 |
+
"step": 2850
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.03,
|
| 412 |
+
"grad_norm": 1.7744035720825195,
|
| 413 |
+
"learning_rate": 0.00012860310421286032,
|
| 414 |
+
"loss": 7.3338,
|
| 415 |
+
"step": 2900
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.03,
|
| 419 |
+
"grad_norm": 2.2014000415802,
|
| 420 |
+
"learning_rate": 0.00013082039911308205,
|
| 421 |
+
"loss": 7.2962,
|
| 422 |
+
"step": 2950
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.03,
|
| 426 |
+
"grad_norm": 1.6716220378875732,
|
| 427 |
+
"learning_rate": 0.00013303769401330377,
|
| 428 |
+
"loss": 7.3348,
|
| 429 |
+
"step": 3000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.03,
|
| 433 |
+
"grad_norm": 1.7045074701309204,
|
| 434 |
+
"learning_rate": 0.0001352549889135255,
|
| 435 |
+
"loss": 7.2864,
|
| 436 |
+
"step": 3050
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.03,
|
| 440 |
+
"grad_norm": 1.8933771848678589,
|
| 441 |
+
"learning_rate": 0.00013747228381374724,
|
| 442 |
+
"loss": 7.2744,
|
| 443 |
+
"step": 3100
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.03,
|
| 447 |
+
"grad_norm": 2.298779249191284,
|
| 448 |
+
"learning_rate": 0.00013968957871396897,
|
| 449 |
+
"loss": 7.2472,
|
| 450 |
+
"step": 3150
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.04,
|
| 454 |
+
"grad_norm": 1.3420922756195068,
|
| 455 |
+
"learning_rate": 0.0001419068736141907,
|
| 456 |
+
"loss": 7.3019,
|
| 457 |
+
"step": 3200
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.04,
|
| 461 |
+
"grad_norm": 1.9339039325714111,
|
| 462 |
+
"learning_rate": 0.00014412416851441242,
|
| 463 |
+
"loss": 7.2982,
|
| 464 |
+
"step": 3250
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.04,
|
| 468 |
+
"grad_norm": 2.69667387008667,
|
| 469 |
+
"learning_rate": 0.00014634146341463414,
|
| 470 |
+
"loss": 7.2851,
|
| 471 |
+
"step": 3300
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.04,
|
| 475 |
+
"grad_norm": 2.3124189376831055,
|
| 476 |
+
"learning_rate": 0.0001485587583148559,
|
| 477 |
+
"loss": 7.258,
|
| 478 |
+
"step": 3350
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.04,
|
| 482 |
+
"grad_norm": 1.975651741027832,
|
| 483 |
+
"learning_rate": 0.00015077605321507762,
|
| 484 |
+
"loss": 7.1275,
|
| 485 |
+
"step": 3400
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.04,
|
| 489 |
+
"grad_norm": 1.9704022407531738,
|
| 490 |
+
"learning_rate": 0.00015299334811529934,
|
| 491 |
+
"loss": 7.1473,
|
| 492 |
+
"step": 3450
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.04,
|
| 496 |
+
"grad_norm": 2.5047757625579834,
|
| 497 |
+
"learning_rate": 0.00015521064301552106,
|
| 498 |
+
"loss": 7.1096,
|
| 499 |
+
"step": 3500
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.04,
|
| 503 |
+
"grad_norm": 1.5465894937515259,
|
| 504 |
+
"learning_rate": 0.0001574279379157428,
|
| 505 |
+
"loss": 7.1501,
|
| 506 |
+
"step": 3550
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.04,
|
| 510 |
+
"grad_norm": 1.9557933807373047,
|
| 511 |
+
"learning_rate": 0.00015964523281596454,
|
| 512 |
+
"loss": 7.2033,
|
| 513 |
+
"step": 3600
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.04,
|
| 517 |
+
"grad_norm": 2.420116424560547,
|
| 518 |
+
"learning_rate": 0.00016186252771618626,
|
| 519 |
+
"loss": 7.1275,
|
| 520 |
+
"step": 3650
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.04,
|
| 524 |
+
"grad_norm": 2.114737033843994,
|
| 525 |
+
"learning_rate": 0.000164079822616408,
|
| 526 |
+
"loss": 7.0932,
|
| 527 |
+
"step": 3700
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.04,
|
| 531 |
+
"grad_norm": 2.3085389137268066,
|
| 532 |
+
"learning_rate": 0.00016629711751662974,
|
| 533 |
+
"loss": 7.0311,
|
| 534 |
+
"step": 3750
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.04,
|
| 538 |
+
"grad_norm": 2.5679140090942383,
|
| 539 |
+
"learning_rate": 0.00016851441241685144,
|
| 540 |
+
"loss": 6.9168,
|
| 541 |
+
"step": 3800
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.04,
|
| 545 |
+
"grad_norm": 1.8611838817596436,
|
| 546 |
+
"learning_rate": 0.0001707317073170732,
|
| 547 |
+
"loss": 7.0085,
|
| 548 |
+
"step": 3850
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.04,
|
| 552 |
+
"grad_norm": 1.8603994846343994,
|
| 553 |
+
"learning_rate": 0.0001729490022172949,
|
| 554 |
+
"loss": 6.9432,
|
| 555 |
+
"step": 3900
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.04,
|
| 559 |
+
"grad_norm": 2.4244627952575684,
|
| 560 |
+
"learning_rate": 0.00017516629711751663,
|
| 561 |
+
"loss": 6.9333,
|
| 562 |
+
"step": 3950
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.04,
|
| 566 |
+
"grad_norm": 2.177870750427246,
|
| 567 |
+
"learning_rate": 0.00017738359201773839,
|
| 568 |
+
"loss": 6.9499,
|
| 569 |
+
"step": 4000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.04,
|
| 573 |
+
"grad_norm": 1.9320554733276367,
|
| 574 |
+
"learning_rate": 0.00017960088691796008,
|
| 575 |
+
"loss": 6.8204,
|
| 576 |
+
"step": 4050
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.05,
|
| 580 |
+
"grad_norm": 1.5062849521636963,
|
| 581 |
+
"learning_rate": 0.00018181818181818183,
|
| 582 |
+
"loss": 6.9505,
|
| 583 |
+
"step": 4100
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.05,
|
| 587 |
+
"grad_norm": 2.9272422790527344,
|
| 588 |
+
"learning_rate": 0.00018403547671840356,
|
| 589 |
+
"loss": 6.8701,
|
| 590 |
+
"step": 4150
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.05,
|
| 594 |
+
"grad_norm": 2.0309596061706543,
|
| 595 |
+
"learning_rate": 0.00018625277161862528,
|
| 596 |
+
"loss": 6.924,
|
| 597 |
+
"step": 4200
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.05,
|
| 601 |
+
"grad_norm": 2.0265886783599854,
|
| 602 |
+
"learning_rate": 0.00018847006651884703,
|
| 603 |
+
"loss": 6.9223,
|
| 604 |
+
"step": 4250
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.05,
|
| 608 |
+
"grad_norm": 2.5160486698150635,
|
| 609 |
+
"learning_rate": 0.00019068736141906876,
|
| 610 |
+
"loss": 6.8708,
|
| 611 |
+
"step": 4300
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.05,
|
| 615 |
+
"grad_norm": 2.613301992416382,
|
| 616 |
+
"learning_rate": 0.00019290465631929045,
|
| 617 |
+
"loss": 6.8937,
|
| 618 |
+
"step": 4350
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.05,
|
| 622 |
+
"grad_norm": 2.3031229972839355,
|
| 623 |
+
"learning_rate": 0.0001951219512195122,
|
| 624 |
+
"loss": 6.8337,
|
| 625 |
+
"step": 4400
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.05,
|
| 629 |
+
"grad_norm": 2.54779052734375,
|
| 630 |
+
"learning_rate": 0.00019733924611973393,
|
| 631 |
+
"loss": 6.8334,
|
| 632 |
+
"step": 4450
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.05,
|
| 636 |
+
"grad_norm": 2.8277971744537354,
|
| 637 |
+
"learning_rate": 0.00019955654101995565,
|
| 638 |
+
"loss": 6.7925,
|
| 639 |
+
"step": 4500
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.05,
|
| 643 |
+
"grad_norm": 2.0113885402679443,
|
| 644 |
+
"learning_rate": 0.00019999989242739025,
|
| 645 |
+
"loss": 6.8458,
|
| 646 |
+
"step": 4550
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.05,
|
| 650 |
+
"grad_norm": 2.2395377159118652,
|
| 651 |
+
"learning_rate": 0.00019999945541405976,
|
| 652 |
+
"loss": 6.6251,
|
| 653 |
+
"step": 4600
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.05,
|
| 657 |
+
"grad_norm": 2.445993423461914,
|
| 658 |
+
"learning_rate": 0.0001999986822381884,
|
| 659 |
+
"loss": 6.8099,
|
| 660 |
+
"step": 4650
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.05,
|
| 664 |
+
"grad_norm": 4.077752590179443,
|
| 665 |
+
"learning_rate": 0.0001999975729023753,
|
| 666 |
+
"loss": 6.8053,
|
| 667 |
+
"step": 4700
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.05,
|
| 671 |
+
"grad_norm": 3.167569875717163,
|
| 672 |
+
"learning_rate": 0.00019999612741034963,
|
| 673 |
+
"loss": 6.7706,
|
| 674 |
+
"step": 4750
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.05,
|
| 678 |
+
"grad_norm": 1.893659234046936,
|
| 679 |
+
"learning_rate": 0.00019999434576697066,
|
| 680 |
+
"loss": 6.8245,
|
| 681 |
+
"step": 4800
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.05,
|
| 685 |
+
"grad_norm": 3.6101326942443848,
|
| 686 |
+
"learning_rate": 0.00019999222797822762,
|
| 687 |
+
"loss": 6.7407,
|
| 688 |
+
"step": 4850
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.05,
|
| 692 |
+
"grad_norm": 2.2858726978302,
|
| 693 |
+
"learning_rate": 0.00019998977405123974,
|
| 694 |
+
"loss": 6.74,
|
| 695 |
+
"step": 4900
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.05,
|
| 699 |
+
"grad_norm": 1.9325459003448486,
|
| 700 |
+
"learning_rate": 0.0001999869839942563,
|
| 701 |
+
"loss": 6.716,
|
| 702 |
+
"step": 4950
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.06,
|
| 706 |
+
"grad_norm": 2.0043437480926514,
|
| 707 |
+
"learning_rate": 0.00019998385781665643,
|
| 708 |
+
"loss": 6.6003,
|
| 709 |
+
"step": 5000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.06,
|
| 713 |
+
"grad_norm": 4.151523113250732,
|
| 714 |
+
"learning_rate": 0.00019998039552894924,
|
| 715 |
+
"loss": 6.6801,
|
| 716 |
+
"step": 5050
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.06,
|
| 720 |
+
"grad_norm": 3.8407771587371826,
|
| 721 |
+
"learning_rate": 0.00019997659714277372,
|
| 722 |
+
"loss": 6.608,
|
| 723 |
+
"step": 5100
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.06,
|
| 727 |
+
"grad_norm": 2.230713129043579,
|
| 728 |
+
"learning_rate": 0.00019997246267089867,
|
| 729 |
+
"loss": 6.6479,
|
| 730 |
+
"step": 5150
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.06,
|
| 734 |
+
"grad_norm": 2.2546942234039307,
|
| 735 |
+
"learning_rate": 0.0001999679921272227,
|
| 736 |
+
"loss": 6.6548,
|
| 737 |
+
"step": 5200
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.06,
|
| 741 |
+
"grad_norm": 3.180986166000366,
|
| 742 |
+
"learning_rate": 0.00019996318552677425,
|
| 743 |
+
"loss": 6.6851,
|
| 744 |
+
"step": 5250
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.06,
|
| 748 |
+
"grad_norm": 2.341231346130371,
|
| 749 |
+
"learning_rate": 0.00019995804288571134,
|
| 750 |
+
"loss": 6.547,
|
| 751 |
+
"step": 5300
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.06,
|
| 755 |
+
"grad_norm": 3.1117124557495117,
|
| 756 |
+
"learning_rate": 0.00019995256422132172,
|
| 757 |
+
"loss": 6.7072,
|
| 758 |
+
"step": 5350
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.06,
|
| 762 |
+
"grad_norm": 2.0082530975341797,
|
| 763 |
+
"learning_rate": 0.0001999467495520227,
|
| 764 |
+
"loss": 6.5422,
|
| 765 |
+
"step": 5400
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.06,
|
| 769 |
+
"grad_norm": 2.409489870071411,
|
| 770 |
+
"learning_rate": 0.0001999405988973611,
|
| 771 |
+
"loss": 6.3716,
|
| 772 |
+
"step": 5450
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.06,
|
| 776 |
+
"grad_norm": 2.649052381515503,
|
| 777 |
+
"learning_rate": 0.00019993411227801328,
|
| 778 |
+
"loss": 6.6434,
|
| 779 |
+
"step": 5500
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.06,
|
| 783 |
+
"grad_norm": 3.081116199493408,
|
| 784 |
+
"learning_rate": 0.00019992728971578492,
|
| 785 |
+
"loss": 6.4624,
|
| 786 |
+
"step": 5550
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.06,
|
| 790 |
+
"grad_norm": 3.1578280925750732,
|
| 791 |
+
"learning_rate": 0.00019992013123361102,
|
| 792 |
+
"loss": 6.5416,
|
| 793 |
+
"step": 5600
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.06,
|
| 797 |
+
"grad_norm": 3.7874557971954346,
|
| 798 |
+
"learning_rate": 0.0001999126368555559,
|
| 799 |
+
"loss": 6.4512,
|
| 800 |
+
"step": 5650
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.06,
|
| 804 |
+
"grad_norm": 2.7693099975585938,
|
| 805 |
+
"learning_rate": 0.00019990480660681293,
|
| 806 |
+
"loss": 6.5105,
|
| 807 |
+
"step": 5700
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.06,
|
| 811 |
+
"grad_norm": 2.4338185787200928,
|
| 812 |
+
"learning_rate": 0.00019989680712666593,
|
| 813 |
+
"loss": 6.5092,
|
| 814 |
+
"step": 5750
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.06,
|
| 818 |
+
"grad_norm": 3.656937837600708,
|
| 819 |
+
"learning_rate": 0.00019988831193270577,
|
| 820 |
+
"loss": 6.4269,
|
| 821 |
+
"step": 5800
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.06,
|
| 825 |
+
"grad_norm": 2.857292652130127,
|
| 826 |
+
"learning_rate": 0.00019987948094982952,
|
| 827 |
+
"loss": 6.4387,
|
| 828 |
+
"step": 5850
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.07,
|
| 832 |
+
"grad_norm": 3.4963467121124268,
|
| 833 |
+
"learning_rate": 0.00019987031420772385,
|
| 834 |
+
"loss": 6.3851,
|
| 835 |
+
"step": 5900
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.07,
|
| 839 |
+
"grad_norm": 2.602522611618042,
|
| 840 |
+
"learning_rate": 0.00019986081173720396,
|
| 841 |
+
"loss": 6.3413,
|
| 842 |
+
"step": 5950
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.07,
|
| 846 |
+
"grad_norm": 2.6455273628234863,
|
| 847 |
+
"learning_rate": 0.00019985097357021385,
|
| 848 |
+
"loss": 6.2965,
|
| 849 |
+
"step": 6000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.07,
|
| 853 |
+
"grad_norm": 3.5592167377471924,
|
| 854 |
+
"learning_rate": 0.0001998407997398259,
|
| 855 |
+
"loss": 6.4293,
|
| 856 |
+
"step": 6050
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.07,
|
| 860 |
+
"grad_norm": 3.6016533374786377,
|
| 861 |
+
"learning_rate": 0.00019983029028024094,
|
| 862 |
+
"loss": 6.2897,
|
| 863 |
+
"step": 6100
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.07,
|
| 867 |
+
"grad_norm": 2.5536839962005615,
|
| 868 |
+
"learning_rate": 0.000199819445226788,
|
| 869 |
+
"loss": 6.3157,
|
| 870 |
+
"step": 6150
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.07,
|
| 874 |
+
"grad_norm": 2.0514349937438965,
|
| 875 |
+
"learning_rate": 0.00019980826461592427,
|
| 876 |
+
"loss": 6.3847,
|
| 877 |
+
"step": 6200
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.07,
|
| 881 |
+
"grad_norm": 2.72495174407959,
|
| 882 |
+
"learning_rate": 0.00019979674848523505,
|
| 883 |
+
"loss": 6.3517,
|
| 884 |
+
"step": 6250
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.07,
|
| 888 |
+
"grad_norm": 2.4264872074127197,
|
| 889 |
+
"learning_rate": 0.00019978489687343335,
|
| 890 |
+
"loss": 6.2533,
|
| 891 |
+
"step": 6300
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.07,
|
| 895 |
+
"grad_norm": 2.8361423015594482,
|
| 896 |
+
"learning_rate": 0.0001997727098203602,
|
| 897 |
+
"loss": 6.3654,
|
| 898 |
+
"step": 6350
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.07,
|
| 902 |
+
"grad_norm": 2.9690892696380615,
|
| 903 |
+
"learning_rate": 0.00019976018736698404,
|
| 904 |
+
"loss": 6.3968,
|
| 905 |
+
"step": 6400
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.07,
|
| 909 |
+
"grad_norm": 2.6132867336273193,
|
| 910 |
+
"learning_rate": 0.0001997473295554009,
|
| 911 |
+
"loss": 6.3444,
|
| 912 |
+
"step": 6450
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.07,
|
| 916 |
+
"grad_norm": 4.820697784423828,
|
| 917 |
+
"learning_rate": 0.00019973413642883424,
|
| 918 |
+
"loss": 6.2019,
|
| 919 |
+
"step": 6500
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.07,
|
| 923 |
+
"grad_norm": 2.2316782474517822,
|
| 924 |
+
"learning_rate": 0.00019972060803163458,
|
| 925 |
+
"loss": 6.2049,
|
| 926 |
+
"step": 6550
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.07,
|
| 930 |
+
"grad_norm": 3.9528305530548096,
|
| 931 |
+
"learning_rate": 0.00019970674440927957,
|
| 932 |
+
"loss": 6.1718,
|
| 933 |
+
"step": 6600
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.07,
|
| 937 |
+
"grad_norm": 1.891073226928711,
|
| 938 |
+
"learning_rate": 0.0001996925456083738,
|
| 939 |
+
"loss": 6.2393,
|
| 940 |
+
"step": 6650
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.07,
|
| 944 |
+
"grad_norm": 2.813270092010498,
|
| 945 |
+
"learning_rate": 0.00019967801167664853,
|
| 946 |
+
"loss": 6.2116,
|
| 947 |
+
"step": 6700
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.07,
|
| 951 |
+
"grad_norm": 2.2726826667785645,
|
| 952 |
+
"learning_rate": 0.00019966314266296173,
|
| 953 |
+
"loss": 6.1521,
|
| 954 |
+
"step": 6750
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.08,
|
| 958 |
+
"grad_norm": 2.3895318508148193,
|
| 959 |
+
"learning_rate": 0.00019964793861729772,
|
| 960 |
+
"loss": 6.1072,
|
| 961 |
+
"step": 6800
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.08,
|
| 965 |
+
"grad_norm": 3.190431833267212,
|
| 966 |
+
"learning_rate": 0.000199632399590767,
|
| 967 |
+
"loss": 6.2009,
|
| 968 |
+
"step": 6850
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.08,
|
| 972 |
+
"grad_norm": 3.79266095161438,
|
| 973 |
+
"learning_rate": 0.00019961652563560634,
|
| 974 |
+
"loss": 6.028,
|
| 975 |
+
"step": 6900
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.08,
|
| 979 |
+
"grad_norm": 3.260039806365967,
|
| 980 |
+
"learning_rate": 0.00019960031680517826,
|
| 981 |
+
"loss": 6.0733,
|
| 982 |
+
"step": 6950
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.08,
|
| 986 |
+
"grad_norm": 3.0739686489105225,
|
| 987 |
+
"learning_rate": 0.0001995837731539711,
|
| 988 |
+
"loss": 6.0521,
|
| 989 |
+
"step": 7000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.08,
|
| 993 |
+
"grad_norm": 3.0517771244049072,
|
| 994 |
+
"learning_rate": 0.00019956689473759872,
|
| 995 |
+
"loss": 6.0544,
|
| 996 |
+
"step": 7050
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.08,
|
| 1000 |
+
"grad_norm": 3.9524648189544678,
|
| 1001 |
+
"learning_rate": 0.0001995496816128003,
|
| 1002 |
+
"loss": 6.1326,
|
| 1003 |
+
"step": 7100
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.08,
|
| 1007 |
+
"grad_norm": 4.498497486114502,
|
| 1008 |
+
"learning_rate": 0.00019953213383744033,
|
| 1009 |
+
"loss": 6.236,
|
| 1010 |
+
"step": 7150
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.08,
|
| 1014 |
+
"grad_norm": 4.157576084136963,
|
| 1015 |
+
"learning_rate": 0.00019951425147050807,
|
| 1016 |
+
"loss": 5.9898,
|
| 1017 |
+
"step": 7200
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.08,
|
| 1021 |
+
"grad_norm": 3.9297516345977783,
|
| 1022 |
+
"learning_rate": 0.00019949603457211775,
|
| 1023 |
+
"loss": 6.086,
|
| 1024 |
+
"step": 7250
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.08,
|
| 1028 |
+
"grad_norm": 3.3214786052703857,
|
| 1029 |
+
"learning_rate": 0.00019947748320350804,
|
| 1030 |
+
"loss": 5.9589,
|
| 1031 |
+
"step": 7300
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.08,
|
| 1035 |
+
"grad_norm": 2.8847291469573975,
|
| 1036 |
+
"learning_rate": 0.00019945859742704201,
|
| 1037 |
+
"loss": 6.1931,
|
| 1038 |
+
"step": 7350
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.08,
|
| 1042 |
+
"grad_norm": 3.387896776199341,
|
| 1043 |
+
"learning_rate": 0.00019943937730620702,
|
| 1044 |
+
"loss": 6.0539,
|
| 1045 |
+
"step": 7400
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.08,
|
| 1049 |
+
"grad_norm": 3.1214797496795654,
|
| 1050 |
+
"learning_rate": 0.00019941982290561417,
|
| 1051 |
+
"loss": 6.0288,
|
| 1052 |
+
"step": 7450
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.08,
|
| 1056 |
+
"grad_norm": 3.7995123863220215,
|
| 1057 |
+
"learning_rate": 0.00019939993429099841,
|
| 1058 |
+
"loss": 6.0526,
|
| 1059 |
+
"step": 7500
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.08,
|
| 1063 |
+
"grad_norm": 4.788393974304199,
|
| 1064 |
+
"learning_rate": 0.00019937971152921818,
|
| 1065 |
+
"loss": 5.9799,
|
| 1066 |
+
"step": 7550
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.08,
|
| 1070 |
+
"grad_norm": 4.009220123291016,
|
| 1071 |
+
"learning_rate": 0.0001993591546882552,
|
| 1072 |
+
"loss": 6.1223,
|
| 1073 |
+
"step": 7600
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.08,
|
| 1077 |
+
"grad_norm": 3.5576276779174805,
|
| 1078 |
+
"learning_rate": 0.00019933826383721428,
|
| 1079 |
+
"loss": 5.989,
|
| 1080 |
+
"step": 7650
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.09,
|
| 1084 |
+
"grad_norm": 3.1287412643432617,
|
| 1085 |
+
"learning_rate": 0.00019931703904632294,
|
| 1086 |
+
"loss": 6.0542,
|
| 1087 |
+
"step": 7700
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.09,
|
| 1091 |
+
"grad_norm": 3.6518595218658447,
|
| 1092 |
+
"learning_rate": 0.00019929548038693146,
|
| 1093 |
+
"loss": 6.041,
|
| 1094 |
+
"step": 7750
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.09,
|
| 1098 |
+
"grad_norm": 3.268080472946167,
|
| 1099 |
+
"learning_rate": 0.0001992735879315123,
|
| 1100 |
+
"loss": 5.888,
|
| 1101 |
+
"step": 7800
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.09,
|
| 1105 |
+
"grad_norm": 3.6055593490600586,
|
| 1106 |
+
"learning_rate": 0.00019925136175366007,
|
| 1107 |
+
"loss": 5.913,
|
| 1108 |
+
"step": 7850
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.09,
|
| 1112 |
+
"grad_norm": 4.866463661193848,
|
| 1113 |
+
"learning_rate": 0.00019922880192809137,
|
| 1114 |
+
"loss": 5.9858,
|
| 1115 |
+
"step": 7900
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.09,
|
| 1119 |
+
"grad_norm": 3.44808292388916,
|
| 1120 |
+
"learning_rate": 0.00019920590853064423,
|
| 1121 |
+
"loss": 5.7686,
|
| 1122 |
+
"step": 7950
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.09,
|
| 1126 |
+
"grad_norm": 2.9507765769958496,
|
| 1127 |
+
"learning_rate": 0.00019918268163827808,
|
| 1128 |
+
"loss": 5.8557,
|
| 1129 |
+
"step": 8000
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.09,
|
| 1133 |
+
"grad_norm": 3.441870927810669,
|
| 1134 |
+
"learning_rate": 0.00019915912132907352,
|
| 1135 |
+
"loss": 5.8268,
|
| 1136 |
+
"step": 8050
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.09,
|
| 1140 |
+
"grad_norm": 3.838809013366699,
|
| 1141 |
+
"learning_rate": 0.00019913522768223182,
|
| 1142 |
+
"loss": 5.9833,
|
| 1143 |
+
"step": 8100
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.09,
|
| 1147 |
+
"grad_norm": 4.165487289428711,
|
| 1148 |
+
"learning_rate": 0.00019911100077807498,
|
| 1149 |
+
"loss": 5.7422,
|
| 1150 |
+
"step": 8150
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.09,
|
| 1154 |
+
"grad_norm": 3.5947463512420654,
|
| 1155 |
+
"learning_rate": 0.0001990864406980452,
|
| 1156 |
+
"loss": 5.7479,
|
| 1157 |
+
"step": 8200
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.09,
|
| 1161 |
+
"grad_norm": 4.130446434020996,
|
| 1162 |
+
"learning_rate": 0.00019906154752470472,
|
| 1163 |
+
"loss": 5.7767,
|
| 1164 |
+
"step": 8250
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.09,
|
| 1168 |
+
"grad_norm": 4.866550922393799,
|
| 1169 |
+
"learning_rate": 0.00019903632134173554,
|
| 1170 |
+
"loss": 5.7681,
|
| 1171 |
+
"step": 8300
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.09,
|
| 1175 |
+
"grad_norm": 3.2839725017547607,
|
| 1176 |
+
"learning_rate": 0.00019901076223393903,
|
| 1177 |
+
"loss": 5.6656,
|
| 1178 |
+
"step": 8350
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.09,
|
| 1182 |
+
"grad_norm": 3.0762476921081543,
|
| 1183 |
+
"learning_rate": 0.0001989848702872359,
|
| 1184 |
+
"loss": 5.789,
|
| 1185 |
+
"step": 8400
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.09,
|
| 1189 |
+
"grad_norm": 3.7109107971191406,
|
| 1190 |
+
"learning_rate": 0.00019895864558866556,
|
| 1191 |
+
"loss": 5.773,
|
| 1192 |
+
"step": 8450
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.09,
|
| 1196 |
+
"grad_norm": 5.400998115539551,
|
| 1197 |
+
"learning_rate": 0.00019893208822638618,
|
| 1198 |
+
"loss": 5.7506,
|
| 1199 |
+
"step": 8500
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.09,
|
| 1203 |
+
"grad_norm": 3.3062849044799805,
|
| 1204 |
+
"learning_rate": 0.00019890519828967413,
|
| 1205 |
+
"loss": 5.7515,
|
| 1206 |
+
"step": 8550
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.1,
|
| 1210 |
+
"grad_norm": 4.109920501708984,
|
| 1211 |
+
"learning_rate": 0.00019887797586892373,
|
| 1212 |
+
"loss": 5.7972,
|
| 1213 |
+
"step": 8600
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.1,
|
| 1217 |
+
"grad_norm": 3.4838390350341797,
|
| 1218 |
+
"learning_rate": 0.00019885042105564717,
|
| 1219 |
+
"loss": 5.6753,
|
| 1220 |
+
"step": 8650
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.1,
|
| 1224 |
+
"grad_norm": 4.251760959625244,
|
| 1225 |
+
"learning_rate": 0.00019882253394247381,
|
| 1226 |
+
"loss": 5.6303,
|
| 1227 |
+
"step": 8700
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.1,
|
| 1231 |
+
"grad_norm": 4.042376518249512,
|
| 1232 |
+
"learning_rate": 0.00019879431462315025,
|
| 1233 |
+
"loss": 5.5753,
|
| 1234 |
+
"step": 8750
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.1,
|
| 1238 |
+
"grad_norm": 4.239652633666992,
|
| 1239 |
+
"learning_rate": 0.0001987657631925398,
|
| 1240 |
+
"loss": 5.5335,
|
| 1241 |
+
"step": 8800
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.1,
|
| 1245 |
+
"grad_norm": 5.15481424331665,
|
| 1246 |
+
"learning_rate": 0.00019873687974662215,
|
| 1247 |
+
"loss": 5.5396,
|
| 1248 |
+
"step": 8850
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 0.1,
|
| 1252 |
+
"grad_norm": 4.36835241317749,
|
| 1253 |
+
"learning_rate": 0.00019870766438249317,
|
| 1254 |
+
"loss": 5.6017,
|
| 1255 |
+
"step": 8900
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.1,
|
| 1259 |
+
"grad_norm": 4.165258407592773,
|
| 1260 |
+
"learning_rate": 0.00019867811719836452,
|
| 1261 |
+
"loss": 5.7228,
|
| 1262 |
+
"step": 8950
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.1,
|
| 1266 |
+
"grad_norm": 4.125988006591797,
|
| 1267 |
+
"learning_rate": 0.0001986482382935633,
|
| 1268 |
+
"loss": 5.5787,
|
| 1269 |
+
"step": 9000
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.1,
|
| 1273 |
+
"grad_norm": 4.177731037139893,
|
| 1274 |
+
"learning_rate": 0.0001986180277685317,
|
| 1275 |
+
"loss": 5.5829,
|
| 1276 |
+
"step": 9050
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.1,
|
| 1280 |
+
"grad_norm": 5.006561279296875,
|
| 1281 |
+
"learning_rate": 0.00019858748572482683,
|
| 1282 |
+
"loss": 5.5466,
|
| 1283 |
+
"step": 9100
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.1,
|
| 1287 |
+
"grad_norm": 4.33070182800293,
|
| 1288 |
+
"learning_rate": 0.00019855661226512007,
|
| 1289 |
+
"loss": 5.5544,
|
| 1290 |
+
"step": 9150
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.1,
|
| 1294 |
+
"grad_norm": 4.358560085296631,
|
| 1295 |
+
"learning_rate": 0.00019852540749319708,
|
| 1296 |
+
"loss": 5.4599,
|
| 1297 |
+
"step": 9200
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 0.1,
|
| 1301 |
+
"grad_norm": 4.536096096038818,
|
| 1302 |
+
"learning_rate": 0.00019849387151395708,
|
| 1303 |
+
"loss": 5.4983,
|
| 1304 |
+
"step": 9250
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.1,
|
| 1308 |
+
"grad_norm": 4.66163444519043,
|
| 1309 |
+
"learning_rate": 0.0001984620044334129,
|
| 1310 |
+
"loss": 5.4097,
|
| 1311 |
+
"step": 9300
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.1,
|
| 1315 |
+
"grad_norm": 4.4319233894348145,
|
| 1316 |
+
"learning_rate": 0.00019842980635869024,
|
| 1317 |
+
"loss": 5.4093,
|
| 1318 |
+
"step": 9350
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 0.1,
|
| 1322 |
+
"grad_norm": 4.98419713973999,
|
| 1323 |
+
"learning_rate": 0.0001983972773980276,
|
| 1324 |
+
"loss": 5.4056,
|
| 1325 |
+
"step": 9400
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.1,
|
| 1329 |
+
"grad_norm": 3.6354339122772217,
|
| 1330 |
+
"learning_rate": 0.0001983644176607757,
|
| 1331 |
+
"loss": 5.3171,
|
| 1332 |
+
"step": 9450
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 0.11,
|
| 1336 |
+
"grad_norm": 4.495342254638672,
|
| 1337 |
+
"learning_rate": 0.00019833122725739736,
|
| 1338 |
+
"loss": 5.4521,
|
| 1339 |
+
"step": 9500
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 0.11,
|
| 1343 |
+
"grad_norm": 4.5558671951293945,
|
| 1344 |
+
"learning_rate": 0.00019829770629946678,
|
| 1345 |
+
"loss": 5.5158,
|
| 1346 |
+
"step": 9550
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.11,
|
| 1350 |
+
"grad_norm": 3.7165732383728027,
|
| 1351 |
+
"learning_rate": 0.00019826385489966957,
|
| 1352 |
+
"loss": 5.301,
|
| 1353 |
+
"step": 9600
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.11,
|
| 1357 |
+
"grad_norm": 6.030915260314941,
|
| 1358 |
+
"learning_rate": 0.00019822967317180204,
|
| 1359 |
+
"loss": 5.3316,
|
| 1360 |
+
"step": 9650
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.11,
|
| 1364 |
+
"grad_norm": 5.385923385620117,
|
| 1365 |
+
"learning_rate": 0.00019819516123077094,
|
| 1366 |
+
"loss": 5.3844,
|
| 1367 |
+
"step": 9700
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.11,
|
| 1371 |
+
"grad_norm": 4.383516788482666,
|
| 1372 |
+
"learning_rate": 0.00019816101926755305,
|
| 1373 |
+
"loss": 5.2995,
|
| 1374 |
+
"step": 9750
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.11,
|
| 1378 |
+
"grad_norm": 4.446406364440918,
|
| 1379 |
+
"learning_rate": 0.00019812585384780055,
|
| 1380 |
+
"loss": 5.386,
|
| 1381 |
+
"step": 9800
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 0.11,
|
| 1385 |
+
"grad_norm": 4.345483303070068,
|
| 1386 |
+
"learning_rate": 0.00019809035856388805,
|
| 1387 |
+
"loss": 5.2815,
|
| 1388 |
+
"step": 9850
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.11,
|
| 1392 |
+
"grad_norm": 4.791261672973633,
|
| 1393 |
+
"learning_rate": 0.00019805453353513813,
|
| 1394 |
+
"loss": 5.3757,
|
| 1395 |
+
"step": 9900
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.11,
|
| 1399 |
+
"grad_norm": 5.622151851654053,
|
| 1400 |
+
"learning_rate": 0.00019801837888198172,
|
| 1401 |
+
"loss": 5.4405,
|
| 1402 |
+
"step": 9950
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.11,
|
| 1406 |
+
"grad_norm": 4.934606075286865,
|
| 1407 |
+
"learning_rate": 0.0001979818947259579,
|
| 1408 |
+
"loss": 5.139,
|
| 1409 |
+
"step": 10000
|
| 1410 |
+
}
|
| 1411 |
+
],
|
| 1412 |
+
"logging_steps": 50,
|
| 1413 |
+
"max_steps": 90183,
|
| 1414 |
+
"num_input_tokens_seen": 0,
|
| 1415 |
+
"num_train_epochs": 1,
|
| 1416 |
+
"save_steps": 1000,
|
| 1417 |
+
"total_flos": 3595905269760000.0,
|
| 1418 |
+
"train_batch_size": 16,
|
| 1419 |
+
"trial_name": null,
|
| 1420 |
+
"trial_params": null
|
| 1421 |
+
}
|
checkpoints/checkpoint-10000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bcdffec292848a9e4baadecd0be02bf6760a1c3238602b1aad355cf1ba43d36e
|
| 3 |
+
size 5048
|
checkpoints/checkpoint-11000/config.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"AlbertForMaskedLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0,
|
| 6 |
+
"bos_token_id": 2,
|
| 7 |
+
"classifier_dropout_prob": 0.1,
|
| 8 |
+
"embedding_size": 128,
|
| 9 |
+
"eos_token_id": 3,
|
| 10 |
+
"hidden_act": "gelu_new",
|
| 11 |
+
"hidden_dropout_prob": 0,
|
| 12 |
+
"hidden_size": 768,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"inner_group_num": 1,
|
| 15 |
+
"intermediate_size": 3072,
|
| 16 |
+
"layer_norm_eps": 1e-12,
|
| 17 |
+
"max_position_embeddings": 512,
|
| 18 |
+
"model_type": "albert",
|
| 19 |
+
"num_attention_heads": 12,
|
| 20 |
+
"num_hidden_groups": 1,
|
| 21 |
+
"num_hidden_layers": 12,
|
| 22 |
+
"pad_token_id": 0,
|
| 23 |
+
"position_embedding_type": "absolute",
|
| 24 |
+
"torch_dtype": "float32",
|
| 25 |
+
"transformers_version": "4.38.2",
|
| 26 |
+
"type_vocab_size": 2,
|
| 27 |
+
"vocab_size": 30000
|
| 28 |
+
}
|
checkpoints/checkpoint-11000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:25348887f65de0d617e799333d76281a2bad79b1426a7936f07eb3c7b668adc2
|
| 3 |
+
size 44890256
|
checkpoints/checkpoint-11000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7a21cc22f410dd970e72c3860cf69fe1d8b19c5e15a1f285e9b1e8b9aa4ad030
|
| 3 |
+
size 89797322
|
checkpoints/checkpoint-11000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:deb8cb1e724bdc9e2e6a27da3e206179ead983df0139090202dbc3fc5277650c
|
| 3 |
+
size 14244
|
checkpoints/checkpoint-11000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4427490ea57450bb98724744f69d9020f2eb9abeb16cbfef6b3d1c81a35b0b83
|
| 3 |
+
size 1064
|
checkpoints/checkpoint-11000/trainer_state.json
ADDED
|
@@ -0,0 +1,1561 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.12197420799929033,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 11000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"grad_norm": 3.2580924034118652,
|
| 14 |
+
"learning_rate": 2.2172949002217296e-06,
|
| 15 |
+
"loss": 10.2933,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.0,
|
| 20 |
+
"grad_norm": 2.4347386360168457,
|
| 21 |
+
"learning_rate": 4.434589800443459e-06,
|
| 22 |
+
"loss": 10.1894,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.0,
|
| 27 |
+
"grad_norm": 2.3895885944366455,
|
| 28 |
+
"learning_rate": 6.651884700665188e-06,
|
| 29 |
+
"loss": 10.1424,
|
| 30 |
+
"step": 150
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.0,
|
| 34 |
+
"grad_norm": 2.129647731781006,
|
| 35 |
+
"learning_rate": 8.869179600886918e-06,
|
| 36 |
+
"loss": 10.0995,
|
| 37 |
+
"step": 200
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.0,
|
| 41 |
+
"grad_norm": 2.3564186096191406,
|
| 42 |
+
"learning_rate": 1.1086474501108649e-05,
|
| 43 |
+
"loss": 10.0479,
|
| 44 |
+
"step": 250
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.0,
|
| 48 |
+
"grad_norm": 1.830551028251648,
|
| 49 |
+
"learning_rate": 1.3303769401330377e-05,
|
| 50 |
+
"loss": 9.9971,
|
| 51 |
+
"step": 300
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.0,
|
| 55 |
+
"grad_norm": 2.1173911094665527,
|
| 56 |
+
"learning_rate": 1.5521064301552106e-05,
|
| 57 |
+
"loss": 9.9201,
|
| 58 |
+
"step": 350
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.0,
|
| 62 |
+
"grad_norm": 1.6636557579040527,
|
| 63 |
+
"learning_rate": 1.7738359201773837e-05,
|
| 64 |
+
"loss": 9.8562,
|
| 65 |
+
"step": 400
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.0,
|
| 69 |
+
"grad_norm": 2.4503839015960693,
|
| 70 |
+
"learning_rate": 1.9955654101995567e-05,
|
| 71 |
+
"loss": 9.7599,
|
| 72 |
+
"step": 450
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.01,
|
| 76 |
+
"grad_norm": 1.822424054145813,
|
| 77 |
+
"learning_rate": 2.2172949002217298e-05,
|
| 78 |
+
"loss": 9.6608,
|
| 79 |
+
"step": 500
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.01,
|
| 83 |
+
"grad_norm": 1.6598998308181763,
|
| 84 |
+
"learning_rate": 2.4390243902439026e-05,
|
| 85 |
+
"loss": 9.55,
|
| 86 |
+
"step": 550
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.01,
|
| 90 |
+
"grad_norm": 1.8471707105636597,
|
| 91 |
+
"learning_rate": 2.6607538802660753e-05,
|
| 92 |
+
"loss": 9.4606,
|
| 93 |
+
"step": 600
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.01,
|
| 97 |
+
"grad_norm": 1.4833533763885498,
|
| 98 |
+
"learning_rate": 2.8824833702882487e-05,
|
| 99 |
+
"loss": 9.3283,
|
| 100 |
+
"step": 650
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.01,
|
| 104 |
+
"grad_norm": 1.688541054725647,
|
| 105 |
+
"learning_rate": 3.104212860310421e-05,
|
| 106 |
+
"loss": 9.2229,
|
| 107 |
+
"step": 700
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.01,
|
| 111 |
+
"grad_norm": 1.6466543674468994,
|
| 112 |
+
"learning_rate": 3.325942350332594e-05,
|
| 113 |
+
"loss": 9.1093,
|
| 114 |
+
"step": 750
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.01,
|
| 118 |
+
"grad_norm": 1.4169293642044067,
|
| 119 |
+
"learning_rate": 3.547671840354767e-05,
|
| 120 |
+
"loss": 8.9703,
|
| 121 |
+
"step": 800
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.01,
|
| 125 |
+
"grad_norm": 1.7079193592071533,
|
| 126 |
+
"learning_rate": 3.7694013303769404e-05,
|
| 127 |
+
"loss": 8.8351,
|
| 128 |
+
"step": 850
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.01,
|
| 132 |
+
"grad_norm": 1.5513204336166382,
|
| 133 |
+
"learning_rate": 3.9911308203991135e-05,
|
| 134 |
+
"loss": 8.7111,
|
| 135 |
+
"step": 900
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.01,
|
| 139 |
+
"grad_norm": 1.485573172569275,
|
| 140 |
+
"learning_rate": 4.212860310421286e-05,
|
| 141 |
+
"loss": 8.5627,
|
| 142 |
+
"step": 950
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.01,
|
| 146 |
+
"grad_norm": 1.511690616607666,
|
| 147 |
+
"learning_rate": 4.4345898004434597e-05,
|
| 148 |
+
"loss": 8.5042,
|
| 149 |
+
"step": 1000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.01,
|
| 153 |
+
"grad_norm": 2.1478614807128906,
|
| 154 |
+
"learning_rate": 4.656319290465632e-05,
|
| 155 |
+
"loss": 8.3287,
|
| 156 |
+
"step": 1050
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.01,
|
| 160 |
+
"grad_norm": 1.4060652256011963,
|
| 161 |
+
"learning_rate": 4.878048780487805e-05,
|
| 162 |
+
"loss": 8.2341,
|
| 163 |
+
"step": 1100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.01,
|
| 167 |
+
"grad_norm": 1.3950035572052002,
|
| 168 |
+
"learning_rate": 5.099778270509978e-05,
|
| 169 |
+
"loss": 8.1277,
|
| 170 |
+
"step": 1150
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.01,
|
| 174 |
+
"grad_norm": 1.5197688341140747,
|
| 175 |
+
"learning_rate": 5.3215077605321506e-05,
|
| 176 |
+
"loss": 8.0311,
|
| 177 |
+
"step": 1200
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.01,
|
| 181 |
+
"grad_norm": 1.3406693935394287,
|
| 182 |
+
"learning_rate": 5.543237250554324e-05,
|
| 183 |
+
"loss": 7.9824,
|
| 184 |
+
"step": 1250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.01,
|
| 188 |
+
"grad_norm": 1.4520119428634644,
|
| 189 |
+
"learning_rate": 5.7649667405764975e-05,
|
| 190 |
+
"loss": 7.9948,
|
| 191 |
+
"step": 1300
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.01,
|
| 195 |
+
"grad_norm": 1.179124116897583,
|
| 196 |
+
"learning_rate": 5.98669623059867e-05,
|
| 197 |
+
"loss": 7.9144,
|
| 198 |
+
"step": 1350
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.02,
|
| 202 |
+
"grad_norm": 1.4039533138275146,
|
| 203 |
+
"learning_rate": 6.208425720620842e-05,
|
| 204 |
+
"loss": 7.8768,
|
| 205 |
+
"step": 1400
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.02,
|
| 209 |
+
"grad_norm": 1.5542700290679932,
|
| 210 |
+
"learning_rate": 6.430155210643016e-05,
|
| 211 |
+
"loss": 7.894,
|
| 212 |
+
"step": 1450
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.02,
|
| 216 |
+
"grad_norm": 1.4150550365447998,
|
| 217 |
+
"learning_rate": 6.651884700665188e-05,
|
| 218 |
+
"loss": 7.8409,
|
| 219 |
+
"step": 1500
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.02,
|
| 223 |
+
"grad_norm": 1.6647827625274658,
|
| 224 |
+
"learning_rate": 6.873614190687362e-05,
|
| 225 |
+
"loss": 7.91,
|
| 226 |
+
"step": 1550
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.02,
|
| 230 |
+
"grad_norm": 1.7795697450637817,
|
| 231 |
+
"learning_rate": 7.095343680709535e-05,
|
| 232 |
+
"loss": 7.8256,
|
| 233 |
+
"step": 1600
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.02,
|
| 237 |
+
"grad_norm": 1.933110237121582,
|
| 238 |
+
"learning_rate": 7.317073170731707e-05,
|
| 239 |
+
"loss": 7.8463,
|
| 240 |
+
"step": 1650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.02,
|
| 244 |
+
"grad_norm": 1.1942570209503174,
|
| 245 |
+
"learning_rate": 7.538802660753881e-05,
|
| 246 |
+
"loss": 7.7827,
|
| 247 |
+
"step": 1700
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.02,
|
| 251 |
+
"grad_norm": 1.6759297847747803,
|
| 252 |
+
"learning_rate": 7.760532150776053e-05,
|
| 253 |
+
"loss": 7.8,
|
| 254 |
+
"step": 1750
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.02,
|
| 258 |
+
"grad_norm": 1.093256950378418,
|
| 259 |
+
"learning_rate": 7.982261640798227e-05,
|
| 260 |
+
"loss": 7.7461,
|
| 261 |
+
"step": 1800
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.02,
|
| 265 |
+
"grad_norm": 1.567872166633606,
|
| 266 |
+
"learning_rate": 8.2039911308204e-05,
|
| 267 |
+
"loss": 7.7338,
|
| 268 |
+
"step": 1850
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.02,
|
| 272 |
+
"grad_norm": 1.3017679452896118,
|
| 273 |
+
"learning_rate": 8.425720620842572e-05,
|
| 274 |
+
"loss": 7.804,
|
| 275 |
+
"step": 1900
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.02,
|
| 279 |
+
"grad_norm": 1.7510960102081299,
|
| 280 |
+
"learning_rate": 8.647450110864746e-05,
|
| 281 |
+
"loss": 7.7405,
|
| 282 |
+
"step": 1950
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.02,
|
| 286 |
+
"grad_norm": 1.7215120792388916,
|
| 287 |
+
"learning_rate": 8.869179600886919e-05,
|
| 288 |
+
"loss": 7.7429,
|
| 289 |
+
"step": 2000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.02,
|
| 293 |
+
"grad_norm": 1.6202715635299683,
|
| 294 |
+
"learning_rate": 9.090909090909092e-05,
|
| 295 |
+
"loss": 7.6588,
|
| 296 |
+
"step": 2050
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.02,
|
| 300 |
+
"grad_norm": 1.5680756568908691,
|
| 301 |
+
"learning_rate": 9.312638580931264e-05,
|
| 302 |
+
"loss": 7.6224,
|
| 303 |
+
"step": 2100
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.02,
|
| 307 |
+
"grad_norm": 1.462240219116211,
|
| 308 |
+
"learning_rate": 9.534368070953438e-05,
|
| 309 |
+
"loss": 7.6851,
|
| 310 |
+
"step": 2150
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.02,
|
| 314 |
+
"grad_norm": 2.2018320560455322,
|
| 315 |
+
"learning_rate": 9.75609756097561e-05,
|
| 316 |
+
"loss": 7.6443,
|
| 317 |
+
"step": 2200
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.02,
|
| 321 |
+
"grad_norm": 1.9520208835601807,
|
| 322 |
+
"learning_rate": 9.977827050997783e-05,
|
| 323 |
+
"loss": 7.6456,
|
| 324 |
+
"step": 2250
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.03,
|
| 328 |
+
"grad_norm": 1.115421175956726,
|
| 329 |
+
"learning_rate": 0.00010199556541019956,
|
| 330 |
+
"loss": 7.5894,
|
| 331 |
+
"step": 2300
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.03,
|
| 335 |
+
"grad_norm": 1.6002250909805298,
|
| 336 |
+
"learning_rate": 0.0001042128603104213,
|
| 337 |
+
"loss": 7.6017,
|
| 338 |
+
"step": 2350
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.03,
|
| 342 |
+
"grad_norm": 1.6516796350479126,
|
| 343 |
+
"learning_rate": 0.00010643015521064301,
|
| 344 |
+
"loss": 7.4548,
|
| 345 |
+
"step": 2400
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.03,
|
| 349 |
+
"grad_norm": 2.2168257236480713,
|
| 350 |
+
"learning_rate": 0.00010864745011086475,
|
| 351 |
+
"loss": 7.5867,
|
| 352 |
+
"step": 2450
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.03,
|
| 356 |
+
"grad_norm": 1.5447593927383423,
|
| 357 |
+
"learning_rate": 0.00011086474501108647,
|
| 358 |
+
"loss": 7.5317,
|
| 359 |
+
"step": 2500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.03,
|
| 363 |
+
"grad_norm": 1.6840906143188477,
|
| 364 |
+
"learning_rate": 0.00011308203991130821,
|
| 365 |
+
"loss": 7.5127,
|
| 366 |
+
"step": 2550
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.03,
|
| 370 |
+
"grad_norm": 1.2965503931045532,
|
| 371 |
+
"learning_rate": 0.00011529933481152995,
|
| 372 |
+
"loss": 7.4911,
|
| 373 |
+
"step": 2600
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.03,
|
| 377 |
+
"grad_norm": 1.643584966659546,
|
| 378 |
+
"learning_rate": 0.00011751662971175166,
|
| 379 |
+
"loss": 7.4416,
|
| 380 |
+
"step": 2650
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.03,
|
| 384 |
+
"grad_norm": 1.5419111251831055,
|
| 385 |
+
"learning_rate": 0.0001197339246119734,
|
| 386 |
+
"loss": 7.4944,
|
| 387 |
+
"step": 2700
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.03,
|
| 391 |
+
"grad_norm": 1.7774205207824707,
|
| 392 |
+
"learning_rate": 0.00012195121951219512,
|
| 393 |
+
"loss": 7.4244,
|
| 394 |
+
"step": 2750
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.03,
|
| 398 |
+
"grad_norm": 2.1709322929382324,
|
| 399 |
+
"learning_rate": 0.00012416851441241685,
|
| 400 |
+
"loss": 7.371,
|
| 401 |
+
"step": 2800
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.03,
|
| 405 |
+
"grad_norm": 1.5503411293029785,
|
| 406 |
+
"learning_rate": 0.0001263858093126386,
|
| 407 |
+
"loss": 7.3031,
|
| 408 |
+
"step": 2850
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.03,
|
| 412 |
+
"grad_norm": 1.7744035720825195,
|
| 413 |
+
"learning_rate": 0.00012860310421286032,
|
| 414 |
+
"loss": 7.3338,
|
| 415 |
+
"step": 2900
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.03,
|
| 419 |
+
"grad_norm": 2.2014000415802,
|
| 420 |
+
"learning_rate": 0.00013082039911308205,
|
| 421 |
+
"loss": 7.2962,
|
| 422 |
+
"step": 2950
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.03,
|
| 426 |
+
"grad_norm": 1.6716220378875732,
|
| 427 |
+
"learning_rate": 0.00013303769401330377,
|
| 428 |
+
"loss": 7.3348,
|
| 429 |
+
"step": 3000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.03,
|
| 433 |
+
"grad_norm": 1.7045074701309204,
|
| 434 |
+
"learning_rate": 0.0001352549889135255,
|
| 435 |
+
"loss": 7.2864,
|
| 436 |
+
"step": 3050
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.03,
|
| 440 |
+
"grad_norm": 1.8933771848678589,
|
| 441 |
+
"learning_rate": 0.00013747228381374724,
|
| 442 |
+
"loss": 7.2744,
|
| 443 |
+
"step": 3100
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.03,
|
| 447 |
+
"grad_norm": 2.298779249191284,
|
| 448 |
+
"learning_rate": 0.00013968957871396897,
|
| 449 |
+
"loss": 7.2472,
|
| 450 |
+
"step": 3150
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.04,
|
| 454 |
+
"grad_norm": 1.3420922756195068,
|
| 455 |
+
"learning_rate": 0.0001419068736141907,
|
| 456 |
+
"loss": 7.3019,
|
| 457 |
+
"step": 3200
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.04,
|
| 461 |
+
"grad_norm": 1.9339039325714111,
|
| 462 |
+
"learning_rate": 0.00014412416851441242,
|
| 463 |
+
"loss": 7.2982,
|
| 464 |
+
"step": 3250
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.04,
|
| 468 |
+
"grad_norm": 2.69667387008667,
|
| 469 |
+
"learning_rate": 0.00014634146341463414,
|
| 470 |
+
"loss": 7.2851,
|
| 471 |
+
"step": 3300
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.04,
|
| 475 |
+
"grad_norm": 2.3124189376831055,
|
| 476 |
+
"learning_rate": 0.0001485587583148559,
|
| 477 |
+
"loss": 7.258,
|
| 478 |
+
"step": 3350
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.04,
|
| 482 |
+
"grad_norm": 1.975651741027832,
|
| 483 |
+
"learning_rate": 0.00015077605321507762,
|
| 484 |
+
"loss": 7.1275,
|
| 485 |
+
"step": 3400
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.04,
|
| 489 |
+
"grad_norm": 1.9704022407531738,
|
| 490 |
+
"learning_rate": 0.00015299334811529934,
|
| 491 |
+
"loss": 7.1473,
|
| 492 |
+
"step": 3450
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.04,
|
| 496 |
+
"grad_norm": 2.5047757625579834,
|
| 497 |
+
"learning_rate": 0.00015521064301552106,
|
| 498 |
+
"loss": 7.1096,
|
| 499 |
+
"step": 3500
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.04,
|
| 503 |
+
"grad_norm": 1.5465894937515259,
|
| 504 |
+
"learning_rate": 0.0001574279379157428,
|
| 505 |
+
"loss": 7.1501,
|
| 506 |
+
"step": 3550
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.04,
|
| 510 |
+
"grad_norm": 1.9557933807373047,
|
| 511 |
+
"learning_rate": 0.00015964523281596454,
|
| 512 |
+
"loss": 7.2033,
|
| 513 |
+
"step": 3600
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.04,
|
| 517 |
+
"grad_norm": 2.420116424560547,
|
| 518 |
+
"learning_rate": 0.00016186252771618626,
|
| 519 |
+
"loss": 7.1275,
|
| 520 |
+
"step": 3650
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.04,
|
| 524 |
+
"grad_norm": 2.114737033843994,
|
| 525 |
+
"learning_rate": 0.000164079822616408,
|
| 526 |
+
"loss": 7.0932,
|
| 527 |
+
"step": 3700
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.04,
|
| 531 |
+
"grad_norm": 2.3085389137268066,
|
| 532 |
+
"learning_rate": 0.00016629711751662974,
|
| 533 |
+
"loss": 7.0311,
|
| 534 |
+
"step": 3750
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.04,
|
| 538 |
+
"grad_norm": 2.5679140090942383,
|
| 539 |
+
"learning_rate": 0.00016851441241685144,
|
| 540 |
+
"loss": 6.9168,
|
| 541 |
+
"step": 3800
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.04,
|
| 545 |
+
"grad_norm": 1.8611838817596436,
|
| 546 |
+
"learning_rate": 0.0001707317073170732,
|
| 547 |
+
"loss": 7.0085,
|
| 548 |
+
"step": 3850
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.04,
|
| 552 |
+
"grad_norm": 1.8603994846343994,
|
| 553 |
+
"learning_rate": 0.0001729490022172949,
|
| 554 |
+
"loss": 6.9432,
|
| 555 |
+
"step": 3900
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.04,
|
| 559 |
+
"grad_norm": 2.4244627952575684,
|
| 560 |
+
"learning_rate": 0.00017516629711751663,
|
| 561 |
+
"loss": 6.9333,
|
| 562 |
+
"step": 3950
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.04,
|
| 566 |
+
"grad_norm": 2.177870750427246,
|
| 567 |
+
"learning_rate": 0.00017738359201773839,
|
| 568 |
+
"loss": 6.9499,
|
| 569 |
+
"step": 4000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.04,
|
| 573 |
+
"grad_norm": 1.9320554733276367,
|
| 574 |
+
"learning_rate": 0.00017960088691796008,
|
| 575 |
+
"loss": 6.8204,
|
| 576 |
+
"step": 4050
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.05,
|
| 580 |
+
"grad_norm": 1.5062849521636963,
|
| 581 |
+
"learning_rate": 0.00018181818181818183,
|
| 582 |
+
"loss": 6.9505,
|
| 583 |
+
"step": 4100
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.05,
|
| 587 |
+
"grad_norm": 2.9272422790527344,
|
| 588 |
+
"learning_rate": 0.00018403547671840356,
|
| 589 |
+
"loss": 6.8701,
|
| 590 |
+
"step": 4150
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.05,
|
| 594 |
+
"grad_norm": 2.0309596061706543,
|
| 595 |
+
"learning_rate": 0.00018625277161862528,
|
| 596 |
+
"loss": 6.924,
|
| 597 |
+
"step": 4200
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.05,
|
| 601 |
+
"grad_norm": 2.0265886783599854,
|
| 602 |
+
"learning_rate": 0.00018847006651884703,
|
| 603 |
+
"loss": 6.9223,
|
| 604 |
+
"step": 4250
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.05,
|
| 608 |
+
"grad_norm": 2.5160486698150635,
|
| 609 |
+
"learning_rate": 0.00019068736141906876,
|
| 610 |
+
"loss": 6.8708,
|
| 611 |
+
"step": 4300
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.05,
|
| 615 |
+
"grad_norm": 2.613301992416382,
|
| 616 |
+
"learning_rate": 0.00019290465631929045,
|
| 617 |
+
"loss": 6.8937,
|
| 618 |
+
"step": 4350
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.05,
|
| 622 |
+
"grad_norm": 2.3031229972839355,
|
| 623 |
+
"learning_rate": 0.0001951219512195122,
|
| 624 |
+
"loss": 6.8337,
|
| 625 |
+
"step": 4400
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.05,
|
| 629 |
+
"grad_norm": 2.54779052734375,
|
| 630 |
+
"learning_rate": 0.00019733924611973393,
|
| 631 |
+
"loss": 6.8334,
|
| 632 |
+
"step": 4450
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.05,
|
| 636 |
+
"grad_norm": 2.8277971744537354,
|
| 637 |
+
"learning_rate": 0.00019955654101995565,
|
| 638 |
+
"loss": 6.7925,
|
| 639 |
+
"step": 4500
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.05,
|
| 643 |
+
"grad_norm": 2.0113885402679443,
|
| 644 |
+
"learning_rate": 0.00019999989242739025,
|
| 645 |
+
"loss": 6.8458,
|
| 646 |
+
"step": 4550
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.05,
|
| 650 |
+
"grad_norm": 2.2395377159118652,
|
| 651 |
+
"learning_rate": 0.00019999945541405976,
|
| 652 |
+
"loss": 6.6251,
|
| 653 |
+
"step": 4600
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.05,
|
| 657 |
+
"grad_norm": 2.445993423461914,
|
| 658 |
+
"learning_rate": 0.0001999986822381884,
|
| 659 |
+
"loss": 6.8099,
|
| 660 |
+
"step": 4650
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.05,
|
| 664 |
+
"grad_norm": 4.077752590179443,
|
| 665 |
+
"learning_rate": 0.0001999975729023753,
|
| 666 |
+
"loss": 6.8053,
|
| 667 |
+
"step": 4700
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.05,
|
| 671 |
+
"grad_norm": 3.167569875717163,
|
| 672 |
+
"learning_rate": 0.00019999612741034963,
|
| 673 |
+
"loss": 6.7706,
|
| 674 |
+
"step": 4750
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.05,
|
| 678 |
+
"grad_norm": 1.893659234046936,
|
| 679 |
+
"learning_rate": 0.00019999434576697066,
|
| 680 |
+
"loss": 6.8245,
|
| 681 |
+
"step": 4800
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.05,
|
| 685 |
+
"grad_norm": 3.6101326942443848,
|
| 686 |
+
"learning_rate": 0.00019999222797822762,
|
| 687 |
+
"loss": 6.7407,
|
| 688 |
+
"step": 4850
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.05,
|
| 692 |
+
"grad_norm": 2.2858726978302,
|
| 693 |
+
"learning_rate": 0.00019998977405123974,
|
| 694 |
+
"loss": 6.74,
|
| 695 |
+
"step": 4900
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.05,
|
| 699 |
+
"grad_norm": 1.9325459003448486,
|
| 700 |
+
"learning_rate": 0.0001999869839942563,
|
| 701 |
+
"loss": 6.716,
|
| 702 |
+
"step": 4950
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.06,
|
| 706 |
+
"grad_norm": 2.0043437480926514,
|
| 707 |
+
"learning_rate": 0.00019998385781665643,
|
| 708 |
+
"loss": 6.6003,
|
| 709 |
+
"step": 5000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.06,
|
| 713 |
+
"grad_norm": 4.151523113250732,
|
| 714 |
+
"learning_rate": 0.00019998039552894924,
|
| 715 |
+
"loss": 6.6801,
|
| 716 |
+
"step": 5050
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.06,
|
| 720 |
+
"grad_norm": 3.8407771587371826,
|
| 721 |
+
"learning_rate": 0.00019997659714277372,
|
| 722 |
+
"loss": 6.608,
|
| 723 |
+
"step": 5100
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.06,
|
| 727 |
+
"grad_norm": 2.230713129043579,
|
| 728 |
+
"learning_rate": 0.00019997246267089867,
|
| 729 |
+
"loss": 6.6479,
|
| 730 |
+
"step": 5150
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.06,
|
| 734 |
+
"grad_norm": 2.2546942234039307,
|
| 735 |
+
"learning_rate": 0.0001999679921272227,
|
| 736 |
+
"loss": 6.6548,
|
| 737 |
+
"step": 5200
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.06,
|
| 741 |
+
"grad_norm": 3.180986166000366,
|
| 742 |
+
"learning_rate": 0.00019996318552677425,
|
| 743 |
+
"loss": 6.6851,
|
| 744 |
+
"step": 5250
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.06,
|
| 748 |
+
"grad_norm": 2.341231346130371,
|
| 749 |
+
"learning_rate": 0.00019995804288571134,
|
| 750 |
+
"loss": 6.547,
|
| 751 |
+
"step": 5300
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.06,
|
| 755 |
+
"grad_norm": 3.1117124557495117,
|
| 756 |
+
"learning_rate": 0.00019995256422132172,
|
| 757 |
+
"loss": 6.7072,
|
| 758 |
+
"step": 5350
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.06,
|
| 762 |
+
"grad_norm": 2.0082530975341797,
|
| 763 |
+
"learning_rate": 0.0001999467495520227,
|
| 764 |
+
"loss": 6.5422,
|
| 765 |
+
"step": 5400
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.06,
|
| 769 |
+
"grad_norm": 2.409489870071411,
|
| 770 |
+
"learning_rate": 0.0001999405988973611,
|
| 771 |
+
"loss": 6.3716,
|
| 772 |
+
"step": 5450
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.06,
|
| 776 |
+
"grad_norm": 2.649052381515503,
|
| 777 |
+
"learning_rate": 0.00019993411227801328,
|
| 778 |
+
"loss": 6.6434,
|
| 779 |
+
"step": 5500
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.06,
|
| 783 |
+
"grad_norm": 3.081116199493408,
|
| 784 |
+
"learning_rate": 0.00019992728971578492,
|
| 785 |
+
"loss": 6.4624,
|
| 786 |
+
"step": 5550
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.06,
|
| 790 |
+
"grad_norm": 3.1578280925750732,
|
| 791 |
+
"learning_rate": 0.00019992013123361102,
|
| 792 |
+
"loss": 6.5416,
|
| 793 |
+
"step": 5600
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.06,
|
| 797 |
+
"grad_norm": 3.7874557971954346,
|
| 798 |
+
"learning_rate": 0.0001999126368555559,
|
| 799 |
+
"loss": 6.4512,
|
| 800 |
+
"step": 5650
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.06,
|
| 804 |
+
"grad_norm": 2.7693099975585938,
|
| 805 |
+
"learning_rate": 0.00019990480660681293,
|
| 806 |
+
"loss": 6.5105,
|
| 807 |
+
"step": 5700
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.06,
|
| 811 |
+
"grad_norm": 2.4338185787200928,
|
| 812 |
+
"learning_rate": 0.00019989680712666593,
|
| 813 |
+
"loss": 6.5092,
|
| 814 |
+
"step": 5750
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.06,
|
| 818 |
+
"grad_norm": 3.656937837600708,
|
| 819 |
+
"learning_rate": 0.00019988831193270577,
|
| 820 |
+
"loss": 6.4269,
|
| 821 |
+
"step": 5800
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.06,
|
| 825 |
+
"grad_norm": 2.857292652130127,
|
| 826 |
+
"learning_rate": 0.00019987948094982952,
|
| 827 |
+
"loss": 6.4387,
|
| 828 |
+
"step": 5850
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.07,
|
| 832 |
+
"grad_norm": 3.4963467121124268,
|
| 833 |
+
"learning_rate": 0.00019987031420772385,
|
| 834 |
+
"loss": 6.3851,
|
| 835 |
+
"step": 5900
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.07,
|
| 839 |
+
"grad_norm": 2.602522611618042,
|
| 840 |
+
"learning_rate": 0.00019986081173720396,
|
| 841 |
+
"loss": 6.3413,
|
| 842 |
+
"step": 5950
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.07,
|
| 846 |
+
"grad_norm": 2.6455273628234863,
|
| 847 |
+
"learning_rate": 0.00019985097357021385,
|
| 848 |
+
"loss": 6.2965,
|
| 849 |
+
"step": 6000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.07,
|
| 853 |
+
"grad_norm": 3.5592167377471924,
|
| 854 |
+
"learning_rate": 0.0001998407997398259,
|
| 855 |
+
"loss": 6.4293,
|
| 856 |
+
"step": 6050
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.07,
|
| 860 |
+
"grad_norm": 3.6016533374786377,
|
| 861 |
+
"learning_rate": 0.00019983029028024094,
|
| 862 |
+
"loss": 6.2897,
|
| 863 |
+
"step": 6100
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.07,
|
| 867 |
+
"grad_norm": 2.5536839962005615,
|
| 868 |
+
"learning_rate": 0.000199819445226788,
|
| 869 |
+
"loss": 6.3157,
|
| 870 |
+
"step": 6150
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.07,
|
| 874 |
+
"grad_norm": 2.0514349937438965,
|
| 875 |
+
"learning_rate": 0.00019980826461592427,
|
| 876 |
+
"loss": 6.3847,
|
| 877 |
+
"step": 6200
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.07,
|
| 881 |
+
"grad_norm": 2.72495174407959,
|
| 882 |
+
"learning_rate": 0.00019979674848523505,
|
| 883 |
+
"loss": 6.3517,
|
| 884 |
+
"step": 6250
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.07,
|
| 888 |
+
"grad_norm": 2.4264872074127197,
|
| 889 |
+
"learning_rate": 0.00019978489687343335,
|
| 890 |
+
"loss": 6.2533,
|
| 891 |
+
"step": 6300
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.07,
|
| 895 |
+
"grad_norm": 2.8361423015594482,
|
| 896 |
+
"learning_rate": 0.0001997727098203602,
|
| 897 |
+
"loss": 6.3654,
|
| 898 |
+
"step": 6350
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.07,
|
| 902 |
+
"grad_norm": 2.9690892696380615,
|
| 903 |
+
"learning_rate": 0.00019976018736698404,
|
| 904 |
+
"loss": 6.3968,
|
| 905 |
+
"step": 6400
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.07,
|
| 909 |
+
"grad_norm": 2.6132867336273193,
|
| 910 |
+
"learning_rate": 0.0001997473295554009,
|
| 911 |
+
"loss": 6.3444,
|
| 912 |
+
"step": 6450
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.07,
|
| 916 |
+
"grad_norm": 4.820697784423828,
|
| 917 |
+
"learning_rate": 0.00019973413642883424,
|
| 918 |
+
"loss": 6.2019,
|
| 919 |
+
"step": 6500
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.07,
|
| 923 |
+
"grad_norm": 2.2316782474517822,
|
| 924 |
+
"learning_rate": 0.00019972060803163458,
|
| 925 |
+
"loss": 6.2049,
|
| 926 |
+
"step": 6550
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.07,
|
| 930 |
+
"grad_norm": 3.9528305530548096,
|
| 931 |
+
"learning_rate": 0.00019970674440927957,
|
| 932 |
+
"loss": 6.1718,
|
| 933 |
+
"step": 6600
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.07,
|
| 937 |
+
"grad_norm": 1.891073226928711,
|
| 938 |
+
"learning_rate": 0.0001996925456083738,
|
| 939 |
+
"loss": 6.2393,
|
| 940 |
+
"step": 6650
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.07,
|
| 944 |
+
"grad_norm": 2.813270092010498,
|
| 945 |
+
"learning_rate": 0.00019967801167664853,
|
| 946 |
+
"loss": 6.2116,
|
| 947 |
+
"step": 6700
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.07,
|
| 951 |
+
"grad_norm": 2.2726826667785645,
|
| 952 |
+
"learning_rate": 0.00019966314266296173,
|
| 953 |
+
"loss": 6.1521,
|
| 954 |
+
"step": 6750
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.08,
|
| 958 |
+
"grad_norm": 2.3895318508148193,
|
| 959 |
+
"learning_rate": 0.00019964793861729772,
|
| 960 |
+
"loss": 6.1072,
|
| 961 |
+
"step": 6800
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.08,
|
| 965 |
+
"grad_norm": 3.190431833267212,
|
| 966 |
+
"learning_rate": 0.000199632399590767,
|
| 967 |
+
"loss": 6.2009,
|
| 968 |
+
"step": 6850
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.08,
|
| 972 |
+
"grad_norm": 3.79266095161438,
|
| 973 |
+
"learning_rate": 0.00019961652563560634,
|
| 974 |
+
"loss": 6.028,
|
| 975 |
+
"step": 6900
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.08,
|
| 979 |
+
"grad_norm": 3.260039806365967,
|
| 980 |
+
"learning_rate": 0.00019960031680517826,
|
| 981 |
+
"loss": 6.0733,
|
| 982 |
+
"step": 6950
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.08,
|
| 986 |
+
"grad_norm": 3.0739686489105225,
|
| 987 |
+
"learning_rate": 0.0001995837731539711,
|
| 988 |
+
"loss": 6.0521,
|
| 989 |
+
"step": 7000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.08,
|
| 993 |
+
"grad_norm": 3.0517771244049072,
|
| 994 |
+
"learning_rate": 0.00019956689473759872,
|
| 995 |
+
"loss": 6.0544,
|
| 996 |
+
"step": 7050
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.08,
|
| 1000 |
+
"grad_norm": 3.9524648189544678,
|
| 1001 |
+
"learning_rate": 0.0001995496816128003,
|
| 1002 |
+
"loss": 6.1326,
|
| 1003 |
+
"step": 7100
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.08,
|
| 1007 |
+
"grad_norm": 4.498497486114502,
|
| 1008 |
+
"learning_rate": 0.00019953213383744033,
|
| 1009 |
+
"loss": 6.236,
|
| 1010 |
+
"step": 7150
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.08,
|
| 1014 |
+
"grad_norm": 4.157576084136963,
|
| 1015 |
+
"learning_rate": 0.00019951425147050807,
|
| 1016 |
+
"loss": 5.9898,
|
| 1017 |
+
"step": 7200
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.08,
|
| 1021 |
+
"grad_norm": 3.9297516345977783,
|
| 1022 |
+
"learning_rate": 0.00019949603457211775,
|
| 1023 |
+
"loss": 6.086,
|
| 1024 |
+
"step": 7250
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.08,
|
| 1028 |
+
"grad_norm": 3.3214786052703857,
|
| 1029 |
+
"learning_rate": 0.00019947748320350804,
|
| 1030 |
+
"loss": 5.9589,
|
| 1031 |
+
"step": 7300
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.08,
|
| 1035 |
+
"grad_norm": 2.8847291469573975,
|
| 1036 |
+
"learning_rate": 0.00019945859742704201,
|
| 1037 |
+
"loss": 6.1931,
|
| 1038 |
+
"step": 7350
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.08,
|
| 1042 |
+
"grad_norm": 3.387896776199341,
|
| 1043 |
+
"learning_rate": 0.00019943937730620702,
|
| 1044 |
+
"loss": 6.0539,
|
| 1045 |
+
"step": 7400
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.08,
|
| 1049 |
+
"grad_norm": 3.1214797496795654,
|
| 1050 |
+
"learning_rate": 0.00019941982290561417,
|
| 1051 |
+
"loss": 6.0288,
|
| 1052 |
+
"step": 7450
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.08,
|
| 1056 |
+
"grad_norm": 3.7995123863220215,
|
| 1057 |
+
"learning_rate": 0.00019939993429099841,
|
| 1058 |
+
"loss": 6.0526,
|
| 1059 |
+
"step": 7500
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.08,
|
| 1063 |
+
"grad_norm": 4.788393974304199,
|
| 1064 |
+
"learning_rate": 0.00019937971152921818,
|
| 1065 |
+
"loss": 5.9799,
|
| 1066 |
+
"step": 7550
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.08,
|
| 1070 |
+
"grad_norm": 4.009220123291016,
|
| 1071 |
+
"learning_rate": 0.0001993591546882552,
|
| 1072 |
+
"loss": 6.1223,
|
| 1073 |
+
"step": 7600
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.08,
|
| 1077 |
+
"grad_norm": 3.5576276779174805,
|
| 1078 |
+
"learning_rate": 0.00019933826383721428,
|
| 1079 |
+
"loss": 5.989,
|
| 1080 |
+
"step": 7650
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.09,
|
| 1084 |
+
"grad_norm": 3.1287412643432617,
|
| 1085 |
+
"learning_rate": 0.00019931703904632294,
|
| 1086 |
+
"loss": 6.0542,
|
| 1087 |
+
"step": 7700
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.09,
|
| 1091 |
+
"grad_norm": 3.6518595218658447,
|
| 1092 |
+
"learning_rate": 0.00019929548038693146,
|
| 1093 |
+
"loss": 6.041,
|
| 1094 |
+
"step": 7750
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.09,
|
| 1098 |
+
"grad_norm": 3.268080472946167,
|
| 1099 |
+
"learning_rate": 0.0001992735879315123,
|
| 1100 |
+
"loss": 5.888,
|
| 1101 |
+
"step": 7800
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.09,
|
| 1105 |
+
"grad_norm": 3.6055593490600586,
|
| 1106 |
+
"learning_rate": 0.00019925136175366007,
|
| 1107 |
+
"loss": 5.913,
|
| 1108 |
+
"step": 7850
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.09,
|
| 1112 |
+
"grad_norm": 4.866463661193848,
|
| 1113 |
+
"learning_rate": 0.00019922880192809137,
|
| 1114 |
+
"loss": 5.9858,
|
| 1115 |
+
"step": 7900
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.09,
|
| 1119 |
+
"grad_norm": 3.44808292388916,
|
| 1120 |
+
"learning_rate": 0.00019920590853064423,
|
| 1121 |
+
"loss": 5.7686,
|
| 1122 |
+
"step": 7950
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.09,
|
| 1126 |
+
"grad_norm": 2.9507765769958496,
|
| 1127 |
+
"learning_rate": 0.00019918268163827808,
|
| 1128 |
+
"loss": 5.8557,
|
| 1129 |
+
"step": 8000
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.09,
|
| 1133 |
+
"grad_norm": 3.441870927810669,
|
| 1134 |
+
"learning_rate": 0.00019915912132907352,
|
| 1135 |
+
"loss": 5.8268,
|
| 1136 |
+
"step": 8050
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.09,
|
| 1140 |
+
"grad_norm": 3.838809013366699,
|
| 1141 |
+
"learning_rate": 0.00019913522768223182,
|
| 1142 |
+
"loss": 5.9833,
|
| 1143 |
+
"step": 8100
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.09,
|
| 1147 |
+
"grad_norm": 4.165487289428711,
|
| 1148 |
+
"learning_rate": 0.00019911100077807498,
|
| 1149 |
+
"loss": 5.7422,
|
| 1150 |
+
"step": 8150
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.09,
|
| 1154 |
+
"grad_norm": 3.5947463512420654,
|
| 1155 |
+
"learning_rate": 0.0001990864406980452,
|
| 1156 |
+
"loss": 5.7479,
|
| 1157 |
+
"step": 8200
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.09,
|
| 1161 |
+
"grad_norm": 4.130446434020996,
|
| 1162 |
+
"learning_rate": 0.00019906154752470472,
|
| 1163 |
+
"loss": 5.7767,
|
| 1164 |
+
"step": 8250
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.09,
|
| 1168 |
+
"grad_norm": 4.866550922393799,
|
| 1169 |
+
"learning_rate": 0.00019903632134173554,
|
| 1170 |
+
"loss": 5.7681,
|
| 1171 |
+
"step": 8300
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.09,
|
| 1175 |
+
"grad_norm": 3.2839725017547607,
|
| 1176 |
+
"learning_rate": 0.00019901076223393903,
|
| 1177 |
+
"loss": 5.6656,
|
| 1178 |
+
"step": 8350
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.09,
|
| 1182 |
+
"grad_norm": 3.0762476921081543,
|
| 1183 |
+
"learning_rate": 0.0001989848702872359,
|
| 1184 |
+
"loss": 5.789,
|
| 1185 |
+
"step": 8400
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.09,
|
| 1189 |
+
"grad_norm": 3.7109107971191406,
|
| 1190 |
+
"learning_rate": 0.00019895864558866556,
|
| 1191 |
+
"loss": 5.773,
|
| 1192 |
+
"step": 8450
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.09,
|
| 1196 |
+
"grad_norm": 5.400998115539551,
|
| 1197 |
+
"learning_rate": 0.00019893208822638618,
|
| 1198 |
+
"loss": 5.7506,
|
| 1199 |
+
"step": 8500
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.09,
|
| 1203 |
+
"grad_norm": 3.3062849044799805,
|
| 1204 |
+
"learning_rate": 0.00019890519828967413,
|
| 1205 |
+
"loss": 5.7515,
|
| 1206 |
+
"step": 8550
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.1,
|
| 1210 |
+
"grad_norm": 4.109920501708984,
|
| 1211 |
+
"learning_rate": 0.00019887797586892373,
|
| 1212 |
+
"loss": 5.7972,
|
| 1213 |
+
"step": 8600
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.1,
|
| 1217 |
+
"grad_norm": 3.4838390350341797,
|
| 1218 |
+
"learning_rate": 0.00019885042105564717,
|
| 1219 |
+
"loss": 5.6753,
|
| 1220 |
+
"step": 8650
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.1,
|
| 1224 |
+
"grad_norm": 4.251760959625244,
|
| 1225 |
+
"learning_rate": 0.00019882253394247381,
|
| 1226 |
+
"loss": 5.6303,
|
| 1227 |
+
"step": 8700
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.1,
|
| 1231 |
+
"grad_norm": 4.042376518249512,
|
| 1232 |
+
"learning_rate": 0.00019879431462315025,
|
| 1233 |
+
"loss": 5.5753,
|
| 1234 |
+
"step": 8750
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.1,
|
| 1238 |
+
"grad_norm": 4.239652633666992,
|
| 1239 |
+
"learning_rate": 0.0001987657631925398,
|
| 1240 |
+
"loss": 5.5335,
|
| 1241 |
+
"step": 8800
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.1,
|
| 1245 |
+
"grad_norm": 5.15481424331665,
|
| 1246 |
+
"learning_rate": 0.00019873687974662215,
|
| 1247 |
+
"loss": 5.5396,
|
| 1248 |
+
"step": 8850
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 0.1,
|
| 1252 |
+
"grad_norm": 4.36835241317749,
|
| 1253 |
+
"learning_rate": 0.00019870766438249317,
|
| 1254 |
+
"loss": 5.6017,
|
| 1255 |
+
"step": 8900
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.1,
|
| 1259 |
+
"grad_norm": 4.165258407592773,
|
| 1260 |
+
"learning_rate": 0.00019867811719836452,
|
| 1261 |
+
"loss": 5.7228,
|
| 1262 |
+
"step": 8950
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.1,
|
| 1266 |
+
"grad_norm": 4.125988006591797,
|
| 1267 |
+
"learning_rate": 0.0001986482382935633,
|
| 1268 |
+
"loss": 5.5787,
|
| 1269 |
+
"step": 9000
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.1,
|
| 1273 |
+
"grad_norm": 4.177731037139893,
|
| 1274 |
+
"learning_rate": 0.0001986180277685317,
|
| 1275 |
+
"loss": 5.5829,
|
| 1276 |
+
"step": 9050
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.1,
|
| 1280 |
+
"grad_norm": 5.006561279296875,
|
| 1281 |
+
"learning_rate": 0.00019858748572482683,
|
| 1282 |
+
"loss": 5.5466,
|
| 1283 |
+
"step": 9100
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.1,
|
| 1287 |
+
"grad_norm": 4.33070182800293,
|
| 1288 |
+
"learning_rate": 0.00019855661226512007,
|
| 1289 |
+
"loss": 5.5544,
|
| 1290 |
+
"step": 9150
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.1,
|
| 1294 |
+
"grad_norm": 4.358560085296631,
|
| 1295 |
+
"learning_rate": 0.00019852540749319708,
|
| 1296 |
+
"loss": 5.4599,
|
| 1297 |
+
"step": 9200
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 0.1,
|
| 1301 |
+
"grad_norm": 4.536096096038818,
|
| 1302 |
+
"learning_rate": 0.00019849387151395708,
|
| 1303 |
+
"loss": 5.4983,
|
| 1304 |
+
"step": 9250
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.1,
|
| 1308 |
+
"grad_norm": 4.66163444519043,
|
| 1309 |
+
"learning_rate": 0.0001984620044334129,
|
| 1310 |
+
"loss": 5.4097,
|
| 1311 |
+
"step": 9300
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.1,
|
| 1315 |
+
"grad_norm": 4.4319233894348145,
|
| 1316 |
+
"learning_rate": 0.00019842980635869024,
|
| 1317 |
+
"loss": 5.4093,
|
| 1318 |
+
"step": 9350
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 0.1,
|
| 1322 |
+
"grad_norm": 4.98419713973999,
|
| 1323 |
+
"learning_rate": 0.0001983972773980276,
|
| 1324 |
+
"loss": 5.4056,
|
| 1325 |
+
"step": 9400
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.1,
|
| 1329 |
+
"grad_norm": 3.6354339122772217,
|
| 1330 |
+
"learning_rate": 0.0001983644176607757,
|
| 1331 |
+
"loss": 5.3171,
|
| 1332 |
+
"step": 9450
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 0.11,
|
| 1336 |
+
"grad_norm": 4.495342254638672,
|
| 1337 |
+
"learning_rate": 0.00019833122725739736,
|
| 1338 |
+
"loss": 5.4521,
|
| 1339 |
+
"step": 9500
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 0.11,
|
| 1343 |
+
"grad_norm": 4.5558671951293945,
|
| 1344 |
+
"learning_rate": 0.00019829770629946678,
|
| 1345 |
+
"loss": 5.5158,
|
| 1346 |
+
"step": 9550
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.11,
|
| 1350 |
+
"grad_norm": 3.7165732383728027,
|
| 1351 |
+
"learning_rate": 0.00019826385489966957,
|
| 1352 |
+
"loss": 5.301,
|
| 1353 |
+
"step": 9600
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.11,
|
| 1357 |
+
"grad_norm": 6.030915260314941,
|
| 1358 |
+
"learning_rate": 0.00019822967317180204,
|
| 1359 |
+
"loss": 5.3316,
|
| 1360 |
+
"step": 9650
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.11,
|
| 1364 |
+
"grad_norm": 5.385923385620117,
|
| 1365 |
+
"learning_rate": 0.00019819516123077094,
|
| 1366 |
+
"loss": 5.3844,
|
| 1367 |
+
"step": 9700
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.11,
|
| 1371 |
+
"grad_norm": 4.383516788482666,
|
| 1372 |
+
"learning_rate": 0.00019816101926755305,
|
| 1373 |
+
"loss": 5.2995,
|
| 1374 |
+
"step": 9750
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.11,
|
| 1378 |
+
"grad_norm": 4.446406364440918,
|
| 1379 |
+
"learning_rate": 0.00019812585384780055,
|
| 1380 |
+
"loss": 5.386,
|
| 1381 |
+
"step": 9800
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 0.11,
|
| 1385 |
+
"grad_norm": 4.345483303070068,
|
| 1386 |
+
"learning_rate": 0.00019809035856388805,
|
| 1387 |
+
"loss": 5.2815,
|
| 1388 |
+
"step": 9850
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.11,
|
| 1392 |
+
"grad_norm": 4.791261672973633,
|
| 1393 |
+
"learning_rate": 0.00019805453353513813,
|
| 1394 |
+
"loss": 5.3757,
|
| 1395 |
+
"step": 9900
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.11,
|
| 1399 |
+
"grad_norm": 5.622151851654053,
|
| 1400 |
+
"learning_rate": 0.00019801837888198172,
|
| 1401 |
+
"loss": 5.4405,
|
| 1402 |
+
"step": 9950
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.11,
|
| 1406 |
+
"grad_norm": 4.934606075286865,
|
| 1407 |
+
"learning_rate": 0.0001979818947259579,
|
| 1408 |
+
"loss": 5.139,
|
| 1409 |
+
"step": 10000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 0.11,
|
| 1413 |
+
"grad_norm": 3.9659693241119385,
|
| 1414 |
+
"learning_rate": 0.0001979450811897134,
|
| 1415 |
+
"loss": 5.1726,
|
| 1416 |
+
"step": 10050
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 0.11,
|
| 1420 |
+
"grad_norm": 5.214992046356201,
|
| 1421 |
+
"learning_rate": 0.00019790793839700226,
|
| 1422 |
+
"loss": 5.2864,
|
| 1423 |
+
"step": 10100
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 0.11,
|
| 1427 |
+
"grad_norm": 4.5359601974487305,
|
| 1428 |
+
"learning_rate": 0.00019787046647268524,
|
| 1429 |
+
"loss": 5.1443,
|
| 1430 |
+
"step": 10150
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.11,
|
| 1434 |
+
"grad_norm": 4.26462984085083,
|
| 1435 |
+
"learning_rate": 0.00019783266554272962,
|
| 1436 |
+
"loss": 5.0597,
|
| 1437 |
+
"step": 10200
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.11,
|
| 1441 |
+
"grad_norm": 5.053945064544678,
|
| 1442 |
+
"learning_rate": 0.00019779453573420873,
|
| 1443 |
+
"loss": 5.2946,
|
| 1444 |
+
"step": 10250
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 0.11,
|
| 1448 |
+
"grad_norm": 6.082211494445801,
|
| 1449 |
+
"learning_rate": 0.00019775607717530127,
|
| 1450 |
+
"loss": 5.2075,
|
| 1451 |
+
"step": 10300
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 0.11,
|
| 1455 |
+
"grad_norm": 4.107390403747559,
|
| 1456 |
+
"learning_rate": 0.00019771728999529132,
|
| 1457 |
+
"loss": 5.1394,
|
| 1458 |
+
"step": 10350
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 0.12,
|
| 1462 |
+
"grad_norm": 4.58411169052124,
|
| 1463 |
+
"learning_rate": 0.00019767817432456752,
|
| 1464 |
+
"loss": 5.1064,
|
| 1465 |
+
"step": 10400
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.12,
|
| 1469 |
+
"grad_norm": 8.38965892791748,
|
| 1470 |
+
"learning_rate": 0.00019763952239228627,
|
| 1471 |
+
"loss": 5.0808,
|
| 1472 |
+
"step": 10450
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.12,
|
| 1476 |
+
"grad_norm": 3.885803699493408,
|
| 1477 |
+
"learning_rate": 0.00019759975669894338,
|
| 1478 |
+
"loss": 5.0664,
|
| 1479 |
+
"step": 10500
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.12,
|
| 1483 |
+
"grad_norm": 4.1605916023254395,
|
| 1484 |
+
"learning_rate": 0.00019755966290999167,
|
| 1485 |
+
"loss": 5.2469,
|
| 1486 |
+
"step": 10550
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 0.12,
|
| 1490 |
+
"grad_norm": 4.821887016296387,
|
| 1491 |
+
"learning_rate": 0.00019751924116021225,
|
| 1492 |
+
"loss": 5.2451,
|
| 1493 |
+
"step": 10600
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 0.12,
|
| 1497 |
+
"grad_norm": 3.865694761276245,
|
| 1498 |
+
"learning_rate": 0.00019747849158548858,
|
| 1499 |
+
"loss": 5.2334,
|
| 1500 |
+
"step": 10650
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 0.12,
|
| 1504 |
+
"grad_norm": 3.640681028366089,
|
| 1505 |
+
"learning_rate": 0.00019743741432280625,
|
| 1506 |
+
"loss": 5.1206,
|
| 1507 |
+
"step": 10700
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.12,
|
| 1511 |
+
"grad_norm": 4.04166316986084,
|
| 1512 |
+
"learning_rate": 0.00019739600951025236,
|
| 1513 |
+
"loss": 5.0059,
|
| 1514 |
+
"step": 10750
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.12,
|
| 1518 |
+
"grad_norm": 4.637605667114258,
|
| 1519 |
+
"learning_rate": 0.00019735427728701516,
|
| 1520 |
+
"loss": 5.0302,
|
| 1521 |
+
"step": 10800
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.12,
|
| 1525 |
+
"grad_norm": 4.08723783493042,
|
| 1526 |
+
"learning_rate": 0.0001973122177933835,
|
| 1527 |
+
"loss": 5.1551,
|
| 1528 |
+
"step": 10850
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 0.12,
|
| 1532 |
+
"grad_norm": 3.7944953441619873,
|
| 1533 |
+
"learning_rate": 0.00019726983117074643,
|
| 1534 |
+
"loss": 5.0665,
|
| 1535 |
+
"step": 10900
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 0.12,
|
| 1539 |
+
"grad_norm": 5.2847371101379395,
|
| 1540 |
+
"learning_rate": 0.00019722711756159266,
|
| 1541 |
+
"loss": 5.2212,
|
| 1542 |
+
"step": 10950
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 0.12,
|
| 1546 |
+
"grad_norm": 4.109150409698486,
|
| 1547 |
+
"learning_rate": 0.00019718407710951012,
|
| 1548 |
+
"loss": 5.2645,
|
| 1549 |
+
"step": 11000
|
| 1550 |
+
}
|
| 1551 |
+
],
|
| 1552 |
+
"logging_steps": 50,
|
| 1553 |
+
"max_steps": 90183,
|
| 1554 |
+
"num_input_tokens_seen": 0,
|
| 1555 |
+
"num_train_epochs": 1,
|
| 1556 |
+
"save_steps": 1000,
|
| 1557 |
+
"total_flos": 3955495796736000.0,
|
| 1558 |
+
"train_batch_size": 16,
|
| 1559 |
+
"trial_name": null,
|
| 1560 |
+
"trial_params": null
|
| 1561 |
+
}
|
checkpoints/checkpoint-11000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bcdffec292848a9e4baadecd0be02bf6760a1c3238602b1aad355cf1ba43d36e
|
| 3 |
+
size 5048
|
checkpoints/checkpoint-12000/config.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"AlbertForMaskedLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0,
|
| 6 |
+
"bos_token_id": 2,
|
| 7 |
+
"classifier_dropout_prob": 0.1,
|
| 8 |
+
"embedding_size": 128,
|
| 9 |
+
"eos_token_id": 3,
|
| 10 |
+
"hidden_act": "gelu_new",
|
| 11 |
+
"hidden_dropout_prob": 0,
|
| 12 |
+
"hidden_size": 768,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"inner_group_num": 1,
|
| 15 |
+
"intermediate_size": 3072,
|
| 16 |
+
"layer_norm_eps": 1e-12,
|
| 17 |
+
"max_position_embeddings": 512,
|
| 18 |
+
"model_type": "albert",
|
| 19 |
+
"num_attention_heads": 12,
|
| 20 |
+
"num_hidden_groups": 1,
|
| 21 |
+
"num_hidden_layers": 12,
|
| 22 |
+
"pad_token_id": 0,
|
| 23 |
+
"position_embedding_type": "absolute",
|
| 24 |
+
"torch_dtype": "float32",
|
| 25 |
+
"transformers_version": "4.38.2",
|
| 26 |
+
"type_vocab_size": 2,
|
| 27 |
+
"vocab_size": 30000
|
| 28 |
+
}
|
checkpoints/checkpoint-12000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0f276e7c0de7c6891ad3152b35b4ce9f618ce4cbe2cdc7c0e7d07b6896e20913
|
| 3 |
+
size 44890256
|
checkpoints/checkpoint-12000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:309beb29179b0e674390af658e4273e0447b1a5a25efd8577b92e5e8b197bf26
|
| 3 |
+
size 89797322
|
checkpoints/checkpoint-12000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ca322fe9154877e26e0f23bedfd73c0e615c76580100ffb7b7816e97ab5dea00
|
| 3 |
+
size 14244
|
checkpoints/checkpoint-12000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dffe601f97167a0b882e203f19948bffc64d5444f75c50c940ee12a7086525f6
|
| 3 |
+
size 1064
|
checkpoints/checkpoint-12000/trainer_state.json
ADDED
|
@@ -0,0 +1,1701 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.13306277236286218,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 12000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"grad_norm": 3.2580924034118652,
|
| 14 |
+
"learning_rate": 2.2172949002217296e-06,
|
| 15 |
+
"loss": 10.2933,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.0,
|
| 20 |
+
"grad_norm": 2.4347386360168457,
|
| 21 |
+
"learning_rate": 4.434589800443459e-06,
|
| 22 |
+
"loss": 10.1894,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.0,
|
| 27 |
+
"grad_norm": 2.3895885944366455,
|
| 28 |
+
"learning_rate": 6.651884700665188e-06,
|
| 29 |
+
"loss": 10.1424,
|
| 30 |
+
"step": 150
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.0,
|
| 34 |
+
"grad_norm": 2.129647731781006,
|
| 35 |
+
"learning_rate": 8.869179600886918e-06,
|
| 36 |
+
"loss": 10.0995,
|
| 37 |
+
"step": 200
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.0,
|
| 41 |
+
"grad_norm": 2.3564186096191406,
|
| 42 |
+
"learning_rate": 1.1086474501108649e-05,
|
| 43 |
+
"loss": 10.0479,
|
| 44 |
+
"step": 250
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.0,
|
| 48 |
+
"grad_norm": 1.830551028251648,
|
| 49 |
+
"learning_rate": 1.3303769401330377e-05,
|
| 50 |
+
"loss": 9.9971,
|
| 51 |
+
"step": 300
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.0,
|
| 55 |
+
"grad_norm": 2.1173911094665527,
|
| 56 |
+
"learning_rate": 1.5521064301552106e-05,
|
| 57 |
+
"loss": 9.9201,
|
| 58 |
+
"step": 350
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.0,
|
| 62 |
+
"grad_norm": 1.6636557579040527,
|
| 63 |
+
"learning_rate": 1.7738359201773837e-05,
|
| 64 |
+
"loss": 9.8562,
|
| 65 |
+
"step": 400
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.0,
|
| 69 |
+
"grad_norm": 2.4503839015960693,
|
| 70 |
+
"learning_rate": 1.9955654101995567e-05,
|
| 71 |
+
"loss": 9.7599,
|
| 72 |
+
"step": 450
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.01,
|
| 76 |
+
"grad_norm": 1.822424054145813,
|
| 77 |
+
"learning_rate": 2.2172949002217298e-05,
|
| 78 |
+
"loss": 9.6608,
|
| 79 |
+
"step": 500
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.01,
|
| 83 |
+
"grad_norm": 1.6598998308181763,
|
| 84 |
+
"learning_rate": 2.4390243902439026e-05,
|
| 85 |
+
"loss": 9.55,
|
| 86 |
+
"step": 550
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.01,
|
| 90 |
+
"grad_norm": 1.8471707105636597,
|
| 91 |
+
"learning_rate": 2.6607538802660753e-05,
|
| 92 |
+
"loss": 9.4606,
|
| 93 |
+
"step": 600
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.01,
|
| 97 |
+
"grad_norm": 1.4833533763885498,
|
| 98 |
+
"learning_rate": 2.8824833702882487e-05,
|
| 99 |
+
"loss": 9.3283,
|
| 100 |
+
"step": 650
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.01,
|
| 104 |
+
"grad_norm": 1.688541054725647,
|
| 105 |
+
"learning_rate": 3.104212860310421e-05,
|
| 106 |
+
"loss": 9.2229,
|
| 107 |
+
"step": 700
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.01,
|
| 111 |
+
"grad_norm": 1.6466543674468994,
|
| 112 |
+
"learning_rate": 3.325942350332594e-05,
|
| 113 |
+
"loss": 9.1093,
|
| 114 |
+
"step": 750
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.01,
|
| 118 |
+
"grad_norm": 1.4169293642044067,
|
| 119 |
+
"learning_rate": 3.547671840354767e-05,
|
| 120 |
+
"loss": 8.9703,
|
| 121 |
+
"step": 800
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.01,
|
| 125 |
+
"grad_norm": 1.7079193592071533,
|
| 126 |
+
"learning_rate": 3.7694013303769404e-05,
|
| 127 |
+
"loss": 8.8351,
|
| 128 |
+
"step": 850
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.01,
|
| 132 |
+
"grad_norm": 1.5513204336166382,
|
| 133 |
+
"learning_rate": 3.9911308203991135e-05,
|
| 134 |
+
"loss": 8.7111,
|
| 135 |
+
"step": 900
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.01,
|
| 139 |
+
"grad_norm": 1.485573172569275,
|
| 140 |
+
"learning_rate": 4.212860310421286e-05,
|
| 141 |
+
"loss": 8.5627,
|
| 142 |
+
"step": 950
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.01,
|
| 146 |
+
"grad_norm": 1.511690616607666,
|
| 147 |
+
"learning_rate": 4.4345898004434597e-05,
|
| 148 |
+
"loss": 8.5042,
|
| 149 |
+
"step": 1000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.01,
|
| 153 |
+
"grad_norm": 2.1478614807128906,
|
| 154 |
+
"learning_rate": 4.656319290465632e-05,
|
| 155 |
+
"loss": 8.3287,
|
| 156 |
+
"step": 1050
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.01,
|
| 160 |
+
"grad_norm": 1.4060652256011963,
|
| 161 |
+
"learning_rate": 4.878048780487805e-05,
|
| 162 |
+
"loss": 8.2341,
|
| 163 |
+
"step": 1100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.01,
|
| 167 |
+
"grad_norm": 1.3950035572052002,
|
| 168 |
+
"learning_rate": 5.099778270509978e-05,
|
| 169 |
+
"loss": 8.1277,
|
| 170 |
+
"step": 1150
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.01,
|
| 174 |
+
"grad_norm": 1.5197688341140747,
|
| 175 |
+
"learning_rate": 5.3215077605321506e-05,
|
| 176 |
+
"loss": 8.0311,
|
| 177 |
+
"step": 1200
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.01,
|
| 181 |
+
"grad_norm": 1.3406693935394287,
|
| 182 |
+
"learning_rate": 5.543237250554324e-05,
|
| 183 |
+
"loss": 7.9824,
|
| 184 |
+
"step": 1250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.01,
|
| 188 |
+
"grad_norm": 1.4520119428634644,
|
| 189 |
+
"learning_rate": 5.7649667405764975e-05,
|
| 190 |
+
"loss": 7.9948,
|
| 191 |
+
"step": 1300
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.01,
|
| 195 |
+
"grad_norm": 1.179124116897583,
|
| 196 |
+
"learning_rate": 5.98669623059867e-05,
|
| 197 |
+
"loss": 7.9144,
|
| 198 |
+
"step": 1350
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.02,
|
| 202 |
+
"grad_norm": 1.4039533138275146,
|
| 203 |
+
"learning_rate": 6.208425720620842e-05,
|
| 204 |
+
"loss": 7.8768,
|
| 205 |
+
"step": 1400
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.02,
|
| 209 |
+
"grad_norm": 1.5542700290679932,
|
| 210 |
+
"learning_rate": 6.430155210643016e-05,
|
| 211 |
+
"loss": 7.894,
|
| 212 |
+
"step": 1450
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.02,
|
| 216 |
+
"grad_norm": 1.4150550365447998,
|
| 217 |
+
"learning_rate": 6.651884700665188e-05,
|
| 218 |
+
"loss": 7.8409,
|
| 219 |
+
"step": 1500
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.02,
|
| 223 |
+
"grad_norm": 1.6647827625274658,
|
| 224 |
+
"learning_rate": 6.873614190687362e-05,
|
| 225 |
+
"loss": 7.91,
|
| 226 |
+
"step": 1550
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.02,
|
| 230 |
+
"grad_norm": 1.7795697450637817,
|
| 231 |
+
"learning_rate": 7.095343680709535e-05,
|
| 232 |
+
"loss": 7.8256,
|
| 233 |
+
"step": 1600
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.02,
|
| 237 |
+
"grad_norm": 1.933110237121582,
|
| 238 |
+
"learning_rate": 7.317073170731707e-05,
|
| 239 |
+
"loss": 7.8463,
|
| 240 |
+
"step": 1650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.02,
|
| 244 |
+
"grad_norm": 1.1942570209503174,
|
| 245 |
+
"learning_rate": 7.538802660753881e-05,
|
| 246 |
+
"loss": 7.7827,
|
| 247 |
+
"step": 1700
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.02,
|
| 251 |
+
"grad_norm": 1.6759297847747803,
|
| 252 |
+
"learning_rate": 7.760532150776053e-05,
|
| 253 |
+
"loss": 7.8,
|
| 254 |
+
"step": 1750
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.02,
|
| 258 |
+
"grad_norm": 1.093256950378418,
|
| 259 |
+
"learning_rate": 7.982261640798227e-05,
|
| 260 |
+
"loss": 7.7461,
|
| 261 |
+
"step": 1800
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.02,
|
| 265 |
+
"grad_norm": 1.567872166633606,
|
| 266 |
+
"learning_rate": 8.2039911308204e-05,
|
| 267 |
+
"loss": 7.7338,
|
| 268 |
+
"step": 1850
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.02,
|
| 272 |
+
"grad_norm": 1.3017679452896118,
|
| 273 |
+
"learning_rate": 8.425720620842572e-05,
|
| 274 |
+
"loss": 7.804,
|
| 275 |
+
"step": 1900
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.02,
|
| 279 |
+
"grad_norm": 1.7510960102081299,
|
| 280 |
+
"learning_rate": 8.647450110864746e-05,
|
| 281 |
+
"loss": 7.7405,
|
| 282 |
+
"step": 1950
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.02,
|
| 286 |
+
"grad_norm": 1.7215120792388916,
|
| 287 |
+
"learning_rate": 8.869179600886919e-05,
|
| 288 |
+
"loss": 7.7429,
|
| 289 |
+
"step": 2000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.02,
|
| 293 |
+
"grad_norm": 1.6202715635299683,
|
| 294 |
+
"learning_rate": 9.090909090909092e-05,
|
| 295 |
+
"loss": 7.6588,
|
| 296 |
+
"step": 2050
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.02,
|
| 300 |
+
"grad_norm": 1.5680756568908691,
|
| 301 |
+
"learning_rate": 9.312638580931264e-05,
|
| 302 |
+
"loss": 7.6224,
|
| 303 |
+
"step": 2100
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.02,
|
| 307 |
+
"grad_norm": 1.462240219116211,
|
| 308 |
+
"learning_rate": 9.534368070953438e-05,
|
| 309 |
+
"loss": 7.6851,
|
| 310 |
+
"step": 2150
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.02,
|
| 314 |
+
"grad_norm": 2.2018320560455322,
|
| 315 |
+
"learning_rate": 9.75609756097561e-05,
|
| 316 |
+
"loss": 7.6443,
|
| 317 |
+
"step": 2200
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.02,
|
| 321 |
+
"grad_norm": 1.9520208835601807,
|
| 322 |
+
"learning_rate": 9.977827050997783e-05,
|
| 323 |
+
"loss": 7.6456,
|
| 324 |
+
"step": 2250
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.03,
|
| 328 |
+
"grad_norm": 1.115421175956726,
|
| 329 |
+
"learning_rate": 0.00010199556541019956,
|
| 330 |
+
"loss": 7.5894,
|
| 331 |
+
"step": 2300
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.03,
|
| 335 |
+
"grad_norm": 1.6002250909805298,
|
| 336 |
+
"learning_rate": 0.0001042128603104213,
|
| 337 |
+
"loss": 7.6017,
|
| 338 |
+
"step": 2350
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.03,
|
| 342 |
+
"grad_norm": 1.6516796350479126,
|
| 343 |
+
"learning_rate": 0.00010643015521064301,
|
| 344 |
+
"loss": 7.4548,
|
| 345 |
+
"step": 2400
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.03,
|
| 349 |
+
"grad_norm": 2.2168257236480713,
|
| 350 |
+
"learning_rate": 0.00010864745011086475,
|
| 351 |
+
"loss": 7.5867,
|
| 352 |
+
"step": 2450
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.03,
|
| 356 |
+
"grad_norm": 1.5447593927383423,
|
| 357 |
+
"learning_rate": 0.00011086474501108647,
|
| 358 |
+
"loss": 7.5317,
|
| 359 |
+
"step": 2500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.03,
|
| 363 |
+
"grad_norm": 1.6840906143188477,
|
| 364 |
+
"learning_rate": 0.00011308203991130821,
|
| 365 |
+
"loss": 7.5127,
|
| 366 |
+
"step": 2550
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.03,
|
| 370 |
+
"grad_norm": 1.2965503931045532,
|
| 371 |
+
"learning_rate": 0.00011529933481152995,
|
| 372 |
+
"loss": 7.4911,
|
| 373 |
+
"step": 2600
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.03,
|
| 377 |
+
"grad_norm": 1.643584966659546,
|
| 378 |
+
"learning_rate": 0.00011751662971175166,
|
| 379 |
+
"loss": 7.4416,
|
| 380 |
+
"step": 2650
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.03,
|
| 384 |
+
"grad_norm": 1.5419111251831055,
|
| 385 |
+
"learning_rate": 0.0001197339246119734,
|
| 386 |
+
"loss": 7.4944,
|
| 387 |
+
"step": 2700
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.03,
|
| 391 |
+
"grad_norm": 1.7774205207824707,
|
| 392 |
+
"learning_rate": 0.00012195121951219512,
|
| 393 |
+
"loss": 7.4244,
|
| 394 |
+
"step": 2750
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.03,
|
| 398 |
+
"grad_norm": 2.1709322929382324,
|
| 399 |
+
"learning_rate": 0.00012416851441241685,
|
| 400 |
+
"loss": 7.371,
|
| 401 |
+
"step": 2800
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.03,
|
| 405 |
+
"grad_norm": 1.5503411293029785,
|
| 406 |
+
"learning_rate": 0.0001263858093126386,
|
| 407 |
+
"loss": 7.3031,
|
| 408 |
+
"step": 2850
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.03,
|
| 412 |
+
"grad_norm": 1.7744035720825195,
|
| 413 |
+
"learning_rate": 0.00012860310421286032,
|
| 414 |
+
"loss": 7.3338,
|
| 415 |
+
"step": 2900
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.03,
|
| 419 |
+
"grad_norm": 2.2014000415802,
|
| 420 |
+
"learning_rate": 0.00013082039911308205,
|
| 421 |
+
"loss": 7.2962,
|
| 422 |
+
"step": 2950
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.03,
|
| 426 |
+
"grad_norm": 1.6716220378875732,
|
| 427 |
+
"learning_rate": 0.00013303769401330377,
|
| 428 |
+
"loss": 7.3348,
|
| 429 |
+
"step": 3000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.03,
|
| 433 |
+
"grad_norm": 1.7045074701309204,
|
| 434 |
+
"learning_rate": 0.0001352549889135255,
|
| 435 |
+
"loss": 7.2864,
|
| 436 |
+
"step": 3050
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.03,
|
| 440 |
+
"grad_norm": 1.8933771848678589,
|
| 441 |
+
"learning_rate": 0.00013747228381374724,
|
| 442 |
+
"loss": 7.2744,
|
| 443 |
+
"step": 3100
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.03,
|
| 447 |
+
"grad_norm": 2.298779249191284,
|
| 448 |
+
"learning_rate": 0.00013968957871396897,
|
| 449 |
+
"loss": 7.2472,
|
| 450 |
+
"step": 3150
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.04,
|
| 454 |
+
"grad_norm": 1.3420922756195068,
|
| 455 |
+
"learning_rate": 0.0001419068736141907,
|
| 456 |
+
"loss": 7.3019,
|
| 457 |
+
"step": 3200
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.04,
|
| 461 |
+
"grad_norm": 1.9339039325714111,
|
| 462 |
+
"learning_rate": 0.00014412416851441242,
|
| 463 |
+
"loss": 7.2982,
|
| 464 |
+
"step": 3250
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.04,
|
| 468 |
+
"grad_norm": 2.69667387008667,
|
| 469 |
+
"learning_rate": 0.00014634146341463414,
|
| 470 |
+
"loss": 7.2851,
|
| 471 |
+
"step": 3300
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.04,
|
| 475 |
+
"grad_norm": 2.3124189376831055,
|
| 476 |
+
"learning_rate": 0.0001485587583148559,
|
| 477 |
+
"loss": 7.258,
|
| 478 |
+
"step": 3350
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.04,
|
| 482 |
+
"grad_norm": 1.975651741027832,
|
| 483 |
+
"learning_rate": 0.00015077605321507762,
|
| 484 |
+
"loss": 7.1275,
|
| 485 |
+
"step": 3400
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.04,
|
| 489 |
+
"grad_norm": 1.9704022407531738,
|
| 490 |
+
"learning_rate": 0.00015299334811529934,
|
| 491 |
+
"loss": 7.1473,
|
| 492 |
+
"step": 3450
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.04,
|
| 496 |
+
"grad_norm": 2.5047757625579834,
|
| 497 |
+
"learning_rate": 0.00015521064301552106,
|
| 498 |
+
"loss": 7.1096,
|
| 499 |
+
"step": 3500
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.04,
|
| 503 |
+
"grad_norm": 1.5465894937515259,
|
| 504 |
+
"learning_rate": 0.0001574279379157428,
|
| 505 |
+
"loss": 7.1501,
|
| 506 |
+
"step": 3550
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.04,
|
| 510 |
+
"grad_norm": 1.9557933807373047,
|
| 511 |
+
"learning_rate": 0.00015964523281596454,
|
| 512 |
+
"loss": 7.2033,
|
| 513 |
+
"step": 3600
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.04,
|
| 517 |
+
"grad_norm": 2.420116424560547,
|
| 518 |
+
"learning_rate": 0.00016186252771618626,
|
| 519 |
+
"loss": 7.1275,
|
| 520 |
+
"step": 3650
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.04,
|
| 524 |
+
"grad_norm": 2.114737033843994,
|
| 525 |
+
"learning_rate": 0.000164079822616408,
|
| 526 |
+
"loss": 7.0932,
|
| 527 |
+
"step": 3700
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.04,
|
| 531 |
+
"grad_norm": 2.3085389137268066,
|
| 532 |
+
"learning_rate": 0.00016629711751662974,
|
| 533 |
+
"loss": 7.0311,
|
| 534 |
+
"step": 3750
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.04,
|
| 538 |
+
"grad_norm": 2.5679140090942383,
|
| 539 |
+
"learning_rate": 0.00016851441241685144,
|
| 540 |
+
"loss": 6.9168,
|
| 541 |
+
"step": 3800
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.04,
|
| 545 |
+
"grad_norm": 1.8611838817596436,
|
| 546 |
+
"learning_rate": 0.0001707317073170732,
|
| 547 |
+
"loss": 7.0085,
|
| 548 |
+
"step": 3850
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.04,
|
| 552 |
+
"grad_norm": 1.8603994846343994,
|
| 553 |
+
"learning_rate": 0.0001729490022172949,
|
| 554 |
+
"loss": 6.9432,
|
| 555 |
+
"step": 3900
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.04,
|
| 559 |
+
"grad_norm": 2.4244627952575684,
|
| 560 |
+
"learning_rate": 0.00017516629711751663,
|
| 561 |
+
"loss": 6.9333,
|
| 562 |
+
"step": 3950
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.04,
|
| 566 |
+
"grad_norm": 2.177870750427246,
|
| 567 |
+
"learning_rate": 0.00017738359201773839,
|
| 568 |
+
"loss": 6.9499,
|
| 569 |
+
"step": 4000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.04,
|
| 573 |
+
"grad_norm": 1.9320554733276367,
|
| 574 |
+
"learning_rate": 0.00017960088691796008,
|
| 575 |
+
"loss": 6.8204,
|
| 576 |
+
"step": 4050
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.05,
|
| 580 |
+
"grad_norm": 1.5062849521636963,
|
| 581 |
+
"learning_rate": 0.00018181818181818183,
|
| 582 |
+
"loss": 6.9505,
|
| 583 |
+
"step": 4100
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.05,
|
| 587 |
+
"grad_norm": 2.9272422790527344,
|
| 588 |
+
"learning_rate": 0.00018403547671840356,
|
| 589 |
+
"loss": 6.8701,
|
| 590 |
+
"step": 4150
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.05,
|
| 594 |
+
"grad_norm": 2.0309596061706543,
|
| 595 |
+
"learning_rate": 0.00018625277161862528,
|
| 596 |
+
"loss": 6.924,
|
| 597 |
+
"step": 4200
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.05,
|
| 601 |
+
"grad_norm": 2.0265886783599854,
|
| 602 |
+
"learning_rate": 0.00018847006651884703,
|
| 603 |
+
"loss": 6.9223,
|
| 604 |
+
"step": 4250
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.05,
|
| 608 |
+
"grad_norm": 2.5160486698150635,
|
| 609 |
+
"learning_rate": 0.00019068736141906876,
|
| 610 |
+
"loss": 6.8708,
|
| 611 |
+
"step": 4300
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.05,
|
| 615 |
+
"grad_norm": 2.613301992416382,
|
| 616 |
+
"learning_rate": 0.00019290465631929045,
|
| 617 |
+
"loss": 6.8937,
|
| 618 |
+
"step": 4350
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.05,
|
| 622 |
+
"grad_norm": 2.3031229972839355,
|
| 623 |
+
"learning_rate": 0.0001951219512195122,
|
| 624 |
+
"loss": 6.8337,
|
| 625 |
+
"step": 4400
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.05,
|
| 629 |
+
"grad_norm": 2.54779052734375,
|
| 630 |
+
"learning_rate": 0.00019733924611973393,
|
| 631 |
+
"loss": 6.8334,
|
| 632 |
+
"step": 4450
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.05,
|
| 636 |
+
"grad_norm": 2.8277971744537354,
|
| 637 |
+
"learning_rate": 0.00019955654101995565,
|
| 638 |
+
"loss": 6.7925,
|
| 639 |
+
"step": 4500
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.05,
|
| 643 |
+
"grad_norm": 2.0113885402679443,
|
| 644 |
+
"learning_rate": 0.00019999989242739025,
|
| 645 |
+
"loss": 6.8458,
|
| 646 |
+
"step": 4550
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.05,
|
| 650 |
+
"grad_norm": 2.2395377159118652,
|
| 651 |
+
"learning_rate": 0.00019999945541405976,
|
| 652 |
+
"loss": 6.6251,
|
| 653 |
+
"step": 4600
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.05,
|
| 657 |
+
"grad_norm": 2.445993423461914,
|
| 658 |
+
"learning_rate": 0.0001999986822381884,
|
| 659 |
+
"loss": 6.8099,
|
| 660 |
+
"step": 4650
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.05,
|
| 664 |
+
"grad_norm": 4.077752590179443,
|
| 665 |
+
"learning_rate": 0.0001999975729023753,
|
| 666 |
+
"loss": 6.8053,
|
| 667 |
+
"step": 4700
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.05,
|
| 671 |
+
"grad_norm": 3.167569875717163,
|
| 672 |
+
"learning_rate": 0.00019999612741034963,
|
| 673 |
+
"loss": 6.7706,
|
| 674 |
+
"step": 4750
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.05,
|
| 678 |
+
"grad_norm": 1.893659234046936,
|
| 679 |
+
"learning_rate": 0.00019999434576697066,
|
| 680 |
+
"loss": 6.8245,
|
| 681 |
+
"step": 4800
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.05,
|
| 685 |
+
"grad_norm": 3.6101326942443848,
|
| 686 |
+
"learning_rate": 0.00019999222797822762,
|
| 687 |
+
"loss": 6.7407,
|
| 688 |
+
"step": 4850
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.05,
|
| 692 |
+
"grad_norm": 2.2858726978302,
|
| 693 |
+
"learning_rate": 0.00019998977405123974,
|
| 694 |
+
"loss": 6.74,
|
| 695 |
+
"step": 4900
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.05,
|
| 699 |
+
"grad_norm": 1.9325459003448486,
|
| 700 |
+
"learning_rate": 0.0001999869839942563,
|
| 701 |
+
"loss": 6.716,
|
| 702 |
+
"step": 4950
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.06,
|
| 706 |
+
"grad_norm": 2.0043437480926514,
|
| 707 |
+
"learning_rate": 0.00019998385781665643,
|
| 708 |
+
"loss": 6.6003,
|
| 709 |
+
"step": 5000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.06,
|
| 713 |
+
"grad_norm": 4.151523113250732,
|
| 714 |
+
"learning_rate": 0.00019998039552894924,
|
| 715 |
+
"loss": 6.6801,
|
| 716 |
+
"step": 5050
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.06,
|
| 720 |
+
"grad_norm": 3.8407771587371826,
|
| 721 |
+
"learning_rate": 0.00019997659714277372,
|
| 722 |
+
"loss": 6.608,
|
| 723 |
+
"step": 5100
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.06,
|
| 727 |
+
"grad_norm": 2.230713129043579,
|
| 728 |
+
"learning_rate": 0.00019997246267089867,
|
| 729 |
+
"loss": 6.6479,
|
| 730 |
+
"step": 5150
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.06,
|
| 734 |
+
"grad_norm": 2.2546942234039307,
|
| 735 |
+
"learning_rate": 0.0001999679921272227,
|
| 736 |
+
"loss": 6.6548,
|
| 737 |
+
"step": 5200
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.06,
|
| 741 |
+
"grad_norm": 3.180986166000366,
|
| 742 |
+
"learning_rate": 0.00019996318552677425,
|
| 743 |
+
"loss": 6.6851,
|
| 744 |
+
"step": 5250
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.06,
|
| 748 |
+
"grad_norm": 2.341231346130371,
|
| 749 |
+
"learning_rate": 0.00019995804288571134,
|
| 750 |
+
"loss": 6.547,
|
| 751 |
+
"step": 5300
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.06,
|
| 755 |
+
"grad_norm": 3.1117124557495117,
|
| 756 |
+
"learning_rate": 0.00019995256422132172,
|
| 757 |
+
"loss": 6.7072,
|
| 758 |
+
"step": 5350
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.06,
|
| 762 |
+
"grad_norm": 2.0082530975341797,
|
| 763 |
+
"learning_rate": 0.0001999467495520227,
|
| 764 |
+
"loss": 6.5422,
|
| 765 |
+
"step": 5400
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.06,
|
| 769 |
+
"grad_norm": 2.409489870071411,
|
| 770 |
+
"learning_rate": 0.0001999405988973611,
|
| 771 |
+
"loss": 6.3716,
|
| 772 |
+
"step": 5450
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.06,
|
| 776 |
+
"grad_norm": 2.649052381515503,
|
| 777 |
+
"learning_rate": 0.00019993411227801328,
|
| 778 |
+
"loss": 6.6434,
|
| 779 |
+
"step": 5500
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.06,
|
| 783 |
+
"grad_norm": 3.081116199493408,
|
| 784 |
+
"learning_rate": 0.00019992728971578492,
|
| 785 |
+
"loss": 6.4624,
|
| 786 |
+
"step": 5550
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.06,
|
| 790 |
+
"grad_norm": 3.1578280925750732,
|
| 791 |
+
"learning_rate": 0.00019992013123361102,
|
| 792 |
+
"loss": 6.5416,
|
| 793 |
+
"step": 5600
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.06,
|
| 797 |
+
"grad_norm": 3.7874557971954346,
|
| 798 |
+
"learning_rate": 0.0001999126368555559,
|
| 799 |
+
"loss": 6.4512,
|
| 800 |
+
"step": 5650
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.06,
|
| 804 |
+
"grad_norm": 2.7693099975585938,
|
| 805 |
+
"learning_rate": 0.00019990480660681293,
|
| 806 |
+
"loss": 6.5105,
|
| 807 |
+
"step": 5700
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.06,
|
| 811 |
+
"grad_norm": 2.4338185787200928,
|
| 812 |
+
"learning_rate": 0.00019989680712666593,
|
| 813 |
+
"loss": 6.5092,
|
| 814 |
+
"step": 5750
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.06,
|
| 818 |
+
"grad_norm": 3.656937837600708,
|
| 819 |
+
"learning_rate": 0.00019988831193270577,
|
| 820 |
+
"loss": 6.4269,
|
| 821 |
+
"step": 5800
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.06,
|
| 825 |
+
"grad_norm": 2.857292652130127,
|
| 826 |
+
"learning_rate": 0.00019987948094982952,
|
| 827 |
+
"loss": 6.4387,
|
| 828 |
+
"step": 5850
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.07,
|
| 832 |
+
"grad_norm": 3.4963467121124268,
|
| 833 |
+
"learning_rate": 0.00019987031420772385,
|
| 834 |
+
"loss": 6.3851,
|
| 835 |
+
"step": 5900
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.07,
|
| 839 |
+
"grad_norm": 2.602522611618042,
|
| 840 |
+
"learning_rate": 0.00019986081173720396,
|
| 841 |
+
"loss": 6.3413,
|
| 842 |
+
"step": 5950
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.07,
|
| 846 |
+
"grad_norm": 2.6455273628234863,
|
| 847 |
+
"learning_rate": 0.00019985097357021385,
|
| 848 |
+
"loss": 6.2965,
|
| 849 |
+
"step": 6000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.07,
|
| 853 |
+
"grad_norm": 3.5592167377471924,
|
| 854 |
+
"learning_rate": 0.0001998407997398259,
|
| 855 |
+
"loss": 6.4293,
|
| 856 |
+
"step": 6050
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.07,
|
| 860 |
+
"grad_norm": 3.6016533374786377,
|
| 861 |
+
"learning_rate": 0.00019983029028024094,
|
| 862 |
+
"loss": 6.2897,
|
| 863 |
+
"step": 6100
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.07,
|
| 867 |
+
"grad_norm": 2.5536839962005615,
|
| 868 |
+
"learning_rate": 0.000199819445226788,
|
| 869 |
+
"loss": 6.3157,
|
| 870 |
+
"step": 6150
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.07,
|
| 874 |
+
"grad_norm": 2.0514349937438965,
|
| 875 |
+
"learning_rate": 0.00019980826461592427,
|
| 876 |
+
"loss": 6.3847,
|
| 877 |
+
"step": 6200
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.07,
|
| 881 |
+
"grad_norm": 2.72495174407959,
|
| 882 |
+
"learning_rate": 0.00019979674848523505,
|
| 883 |
+
"loss": 6.3517,
|
| 884 |
+
"step": 6250
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.07,
|
| 888 |
+
"grad_norm": 2.4264872074127197,
|
| 889 |
+
"learning_rate": 0.00019978489687343335,
|
| 890 |
+
"loss": 6.2533,
|
| 891 |
+
"step": 6300
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.07,
|
| 895 |
+
"grad_norm": 2.8361423015594482,
|
| 896 |
+
"learning_rate": 0.0001997727098203602,
|
| 897 |
+
"loss": 6.3654,
|
| 898 |
+
"step": 6350
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.07,
|
| 902 |
+
"grad_norm": 2.9690892696380615,
|
| 903 |
+
"learning_rate": 0.00019976018736698404,
|
| 904 |
+
"loss": 6.3968,
|
| 905 |
+
"step": 6400
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.07,
|
| 909 |
+
"grad_norm": 2.6132867336273193,
|
| 910 |
+
"learning_rate": 0.0001997473295554009,
|
| 911 |
+
"loss": 6.3444,
|
| 912 |
+
"step": 6450
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.07,
|
| 916 |
+
"grad_norm": 4.820697784423828,
|
| 917 |
+
"learning_rate": 0.00019973413642883424,
|
| 918 |
+
"loss": 6.2019,
|
| 919 |
+
"step": 6500
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.07,
|
| 923 |
+
"grad_norm": 2.2316782474517822,
|
| 924 |
+
"learning_rate": 0.00019972060803163458,
|
| 925 |
+
"loss": 6.2049,
|
| 926 |
+
"step": 6550
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.07,
|
| 930 |
+
"grad_norm": 3.9528305530548096,
|
| 931 |
+
"learning_rate": 0.00019970674440927957,
|
| 932 |
+
"loss": 6.1718,
|
| 933 |
+
"step": 6600
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.07,
|
| 937 |
+
"grad_norm": 1.891073226928711,
|
| 938 |
+
"learning_rate": 0.0001996925456083738,
|
| 939 |
+
"loss": 6.2393,
|
| 940 |
+
"step": 6650
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.07,
|
| 944 |
+
"grad_norm": 2.813270092010498,
|
| 945 |
+
"learning_rate": 0.00019967801167664853,
|
| 946 |
+
"loss": 6.2116,
|
| 947 |
+
"step": 6700
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.07,
|
| 951 |
+
"grad_norm": 2.2726826667785645,
|
| 952 |
+
"learning_rate": 0.00019966314266296173,
|
| 953 |
+
"loss": 6.1521,
|
| 954 |
+
"step": 6750
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.08,
|
| 958 |
+
"grad_norm": 2.3895318508148193,
|
| 959 |
+
"learning_rate": 0.00019964793861729772,
|
| 960 |
+
"loss": 6.1072,
|
| 961 |
+
"step": 6800
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.08,
|
| 965 |
+
"grad_norm": 3.190431833267212,
|
| 966 |
+
"learning_rate": 0.000199632399590767,
|
| 967 |
+
"loss": 6.2009,
|
| 968 |
+
"step": 6850
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.08,
|
| 972 |
+
"grad_norm": 3.79266095161438,
|
| 973 |
+
"learning_rate": 0.00019961652563560634,
|
| 974 |
+
"loss": 6.028,
|
| 975 |
+
"step": 6900
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.08,
|
| 979 |
+
"grad_norm": 3.260039806365967,
|
| 980 |
+
"learning_rate": 0.00019960031680517826,
|
| 981 |
+
"loss": 6.0733,
|
| 982 |
+
"step": 6950
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.08,
|
| 986 |
+
"grad_norm": 3.0739686489105225,
|
| 987 |
+
"learning_rate": 0.0001995837731539711,
|
| 988 |
+
"loss": 6.0521,
|
| 989 |
+
"step": 7000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.08,
|
| 993 |
+
"grad_norm": 3.0517771244049072,
|
| 994 |
+
"learning_rate": 0.00019956689473759872,
|
| 995 |
+
"loss": 6.0544,
|
| 996 |
+
"step": 7050
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.08,
|
| 1000 |
+
"grad_norm": 3.9524648189544678,
|
| 1001 |
+
"learning_rate": 0.0001995496816128003,
|
| 1002 |
+
"loss": 6.1326,
|
| 1003 |
+
"step": 7100
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.08,
|
| 1007 |
+
"grad_norm": 4.498497486114502,
|
| 1008 |
+
"learning_rate": 0.00019953213383744033,
|
| 1009 |
+
"loss": 6.236,
|
| 1010 |
+
"step": 7150
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.08,
|
| 1014 |
+
"grad_norm": 4.157576084136963,
|
| 1015 |
+
"learning_rate": 0.00019951425147050807,
|
| 1016 |
+
"loss": 5.9898,
|
| 1017 |
+
"step": 7200
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.08,
|
| 1021 |
+
"grad_norm": 3.9297516345977783,
|
| 1022 |
+
"learning_rate": 0.00019949603457211775,
|
| 1023 |
+
"loss": 6.086,
|
| 1024 |
+
"step": 7250
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.08,
|
| 1028 |
+
"grad_norm": 3.3214786052703857,
|
| 1029 |
+
"learning_rate": 0.00019947748320350804,
|
| 1030 |
+
"loss": 5.9589,
|
| 1031 |
+
"step": 7300
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.08,
|
| 1035 |
+
"grad_norm": 2.8847291469573975,
|
| 1036 |
+
"learning_rate": 0.00019945859742704201,
|
| 1037 |
+
"loss": 6.1931,
|
| 1038 |
+
"step": 7350
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.08,
|
| 1042 |
+
"grad_norm": 3.387896776199341,
|
| 1043 |
+
"learning_rate": 0.00019943937730620702,
|
| 1044 |
+
"loss": 6.0539,
|
| 1045 |
+
"step": 7400
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.08,
|
| 1049 |
+
"grad_norm": 3.1214797496795654,
|
| 1050 |
+
"learning_rate": 0.00019941982290561417,
|
| 1051 |
+
"loss": 6.0288,
|
| 1052 |
+
"step": 7450
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.08,
|
| 1056 |
+
"grad_norm": 3.7995123863220215,
|
| 1057 |
+
"learning_rate": 0.00019939993429099841,
|
| 1058 |
+
"loss": 6.0526,
|
| 1059 |
+
"step": 7500
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.08,
|
| 1063 |
+
"grad_norm": 4.788393974304199,
|
| 1064 |
+
"learning_rate": 0.00019937971152921818,
|
| 1065 |
+
"loss": 5.9799,
|
| 1066 |
+
"step": 7550
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.08,
|
| 1070 |
+
"grad_norm": 4.009220123291016,
|
| 1071 |
+
"learning_rate": 0.0001993591546882552,
|
| 1072 |
+
"loss": 6.1223,
|
| 1073 |
+
"step": 7600
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.08,
|
| 1077 |
+
"grad_norm": 3.5576276779174805,
|
| 1078 |
+
"learning_rate": 0.00019933826383721428,
|
| 1079 |
+
"loss": 5.989,
|
| 1080 |
+
"step": 7650
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.09,
|
| 1084 |
+
"grad_norm": 3.1287412643432617,
|
| 1085 |
+
"learning_rate": 0.00019931703904632294,
|
| 1086 |
+
"loss": 6.0542,
|
| 1087 |
+
"step": 7700
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.09,
|
| 1091 |
+
"grad_norm": 3.6518595218658447,
|
| 1092 |
+
"learning_rate": 0.00019929548038693146,
|
| 1093 |
+
"loss": 6.041,
|
| 1094 |
+
"step": 7750
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.09,
|
| 1098 |
+
"grad_norm": 3.268080472946167,
|
| 1099 |
+
"learning_rate": 0.0001992735879315123,
|
| 1100 |
+
"loss": 5.888,
|
| 1101 |
+
"step": 7800
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.09,
|
| 1105 |
+
"grad_norm": 3.6055593490600586,
|
| 1106 |
+
"learning_rate": 0.00019925136175366007,
|
| 1107 |
+
"loss": 5.913,
|
| 1108 |
+
"step": 7850
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.09,
|
| 1112 |
+
"grad_norm": 4.866463661193848,
|
| 1113 |
+
"learning_rate": 0.00019922880192809137,
|
| 1114 |
+
"loss": 5.9858,
|
| 1115 |
+
"step": 7900
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.09,
|
| 1119 |
+
"grad_norm": 3.44808292388916,
|
| 1120 |
+
"learning_rate": 0.00019920590853064423,
|
| 1121 |
+
"loss": 5.7686,
|
| 1122 |
+
"step": 7950
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.09,
|
| 1126 |
+
"grad_norm": 2.9507765769958496,
|
| 1127 |
+
"learning_rate": 0.00019918268163827808,
|
| 1128 |
+
"loss": 5.8557,
|
| 1129 |
+
"step": 8000
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.09,
|
| 1133 |
+
"grad_norm": 3.441870927810669,
|
| 1134 |
+
"learning_rate": 0.00019915912132907352,
|
| 1135 |
+
"loss": 5.8268,
|
| 1136 |
+
"step": 8050
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.09,
|
| 1140 |
+
"grad_norm": 3.838809013366699,
|
| 1141 |
+
"learning_rate": 0.00019913522768223182,
|
| 1142 |
+
"loss": 5.9833,
|
| 1143 |
+
"step": 8100
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.09,
|
| 1147 |
+
"grad_norm": 4.165487289428711,
|
| 1148 |
+
"learning_rate": 0.00019911100077807498,
|
| 1149 |
+
"loss": 5.7422,
|
| 1150 |
+
"step": 8150
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.09,
|
| 1154 |
+
"grad_norm": 3.5947463512420654,
|
| 1155 |
+
"learning_rate": 0.0001990864406980452,
|
| 1156 |
+
"loss": 5.7479,
|
| 1157 |
+
"step": 8200
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.09,
|
| 1161 |
+
"grad_norm": 4.130446434020996,
|
| 1162 |
+
"learning_rate": 0.00019906154752470472,
|
| 1163 |
+
"loss": 5.7767,
|
| 1164 |
+
"step": 8250
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.09,
|
| 1168 |
+
"grad_norm": 4.866550922393799,
|
| 1169 |
+
"learning_rate": 0.00019903632134173554,
|
| 1170 |
+
"loss": 5.7681,
|
| 1171 |
+
"step": 8300
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.09,
|
| 1175 |
+
"grad_norm": 3.2839725017547607,
|
| 1176 |
+
"learning_rate": 0.00019901076223393903,
|
| 1177 |
+
"loss": 5.6656,
|
| 1178 |
+
"step": 8350
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.09,
|
| 1182 |
+
"grad_norm": 3.0762476921081543,
|
| 1183 |
+
"learning_rate": 0.0001989848702872359,
|
| 1184 |
+
"loss": 5.789,
|
| 1185 |
+
"step": 8400
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.09,
|
| 1189 |
+
"grad_norm": 3.7109107971191406,
|
| 1190 |
+
"learning_rate": 0.00019895864558866556,
|
| 1191 |
+
"loss": 5.773,
|
| 1192 |
+
"step": 8450
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.09,
|
| 1196 |
+
"grad_norm": 5.400998115539551,
|
| 1197 |
+
"learning_rate": 0.00019893208822638618,
|
| 1198 |
+
"loss": 5.7506,
|
| 1199 |
+
"step": 8500
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.09,
|
| 1203 |
+
"grad_norm": 3.3062849044799805,
|
| 1204 |
+
"learning_rate": 0.00019890519828967413,
|
| 1205 |
+
"loss": 5.7515,
|
| 1206 |
+
"step": 8550
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.1,
|
| 1210 |
+
"grad_norm": 4.109920501708984,
|
| 1211 |
+
"learning_rate": 0.00019887797586892373,
|
| 1212 |
+
"loss": 5.7972,
|
| 1213 |
+
"step": 8600
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.1,
|
| 1217 |
+
"grad_norm": 3.4838390350341797,
|
| 1218 |
+
"learning_rate": 0.00019885042105564717,
|
| 1219 |
+
"loss": 5.6753,
|
| 1220 |
+
"step": 8650
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.1,
|
| 1224 |
+
"grad_norm": 4.251760959625244,
|
| 1225 |
+
"learning_rate": 0.00019882253394247381,
|
| 1226 |
+
"loss": 5.6303,
|
| 1227 |
+
"step": 8700
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.1,
|
| 1231 |
+
"grad_norm": 4.042376518249512,
|
| 1232 |
+
"learning_rate": 0.00019879431462315025,
|
| 1233 |
+
"loss": 5.5753,
|
| 1234 |
+
"step": 8750
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.1,
|
| 1238 |
+
"grad_norm": 4.239652633666992,
|
| 1239 |
+
"learning_rate": 0.0001987657631925398,
|
| 1240 |
+
"loss": 5.5335,
|
| 1241 |
+
"step": 8800
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.1,
|
| 1245 |
+
"grad_norm": 5.15481424331665,
|
| 1246 |
+
"learning_rate": 0.00019873687974662215,
|
| 1247 |
+
"loss": 5.5396,
|
| 1248 |
+
"step": 8850
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 0.1,
|
| 1252 |
+
"grad_norm": 4.36835241317749,
|
| 1253 |
+
"learning_rate": 0.00019870766438249317,
|
| 1254 |
+
"loss": 5.6017,
|
| 1255 |
+
"step": 8900
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.1,
|
| 1259 |
+
"grad_norm": 4.165258407592773,
|
| 1260 |
+
"learning_rate": 0.00019867811719836452,
|
| 1261 |
+
"loss": 5.7228,
|
| 1262 |
+
"step": 8950
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.1,
|
| 1266 |
+
"grad_norm": 4.125988006591797,
|
| 1267 |
+
"learning_rate": 0.0001986482382935633,
|
| 1268 |
+
"loss": 5.5787,
|
| 1269 |
+
"step": 9000
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.1,
|
| 1273 |
+
"grad_norm": 4.177731037139893,
|
| 1274 |
+
"learning_rate": 0.0001986180277685317,
|
| 1275 |
+
"loss": 5.5829,
|
| 1276 |
+
"step": 9050
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.1,
|
| 1280 |
+
"grad_norm": 5.006561279296875,
|
| 1281 |
+
"learning_rate": 0.00019858748572482683,
|
| 1282 |
+
"loss": 5.5466,
|
| 1283 |
+
"step": 9100
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.1,
|
| 1287 |
+
"grad_norm": 4.33070182800293,
|
| 1288 |
+
"learning_rate": 0.00019855661226512007,
|
| 1289 |
+
"loss": 5.5544,
|
| 1290 |
+
"step": 9150
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.1,
|
| 1294 |
+
"grad_norm": 4.358560085296631,
|
| 1295 |
+
"learning_rate": 0.00019852540749319708,
|
| 1296 |
+
"loss": 5.4599,
|
| 1297 |
+
"step": 9200
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 0.1,
|
| 1301 |
+
"grad_norm": 4.536096096038818,
|
| 1302 |
+
"learning_rate": 0.00019849387151395708,
|
| 1303 |
+
"loss": 5.4983,
|
| 1304 |
+
"step": 9250
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.1,
|
| 1308 |
+
"grad_norm": 4.66163444519043,
|
| 1309 |
+
"learning_rate": 0.0001984620044334129,
|
| 1310 |
+
"loss": 5.4097,
|
| 1311 |
+
"step": 9300
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.1,
|
| 1315 |
+
"grad_norm": 4.4319233894348145,
|
| 1316 |
+
"learning_rate": 0.00019842980635869024,
|
| 1317 |
+
"loss": 5.4093,
|
| 1318 |
+
"step": 9350
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 0.1,
|
| 1322 |
+
"grad_norm": 4.98419713973999,
|
| 1323 |
+
"learning_rate": 0.0001983972773980276,
|
| 1324 |
+
"loss": 5.4056,
|
| 1325 |
+
"step": 9400
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.1,
|
| 1329 |
+
"grad_norm": 3.6354339122772217,
|
| 1330 |
+
"learning_rate": 0.0001983644176607757,
|
| 1331 |
+
"loss": 5.3171,
|
| 1332 |
+
"step": 9450
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 0.11,
|
| 1336 |
+
"grad_norm": 4.495342254638672,
|
| 1337 |
+
"learning_rate": 0.00019833122725739736,
|
| 1338 |
+
"loss": 5.4521,
|
| 1339 |
+
"step": 9500
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 0.11,
|
| 1343 |
+
"grad_norm": 4.5558671951293945,
|
| 1344 |
+
"learning_rate": 0.00019829770629946678,
|
| 1345 |
+
"loss": 5.5158,
|
| 1346 |
+
"step": 9550
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.11,
|
| 1350 |
+
"grad_norm": 3.7165732383728027,
|
| 1351 |
+
"learning_rate": 0.00019826385489966957,
|
| 1352 |
+
"loss": 5.301,
|
| 1353 |
+
"step": 9600
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.11,
|
| 1357 |
+
"grad_norm": 6.030915260314941,
|
| 1358 |
+
"learning_rate": 0.00019822967317180204,
|
| 1359 |
+
"loss": 5.3316,
|
| 1360 |
+
"step": 9650
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.11,
|
| 1364 |
+
"grad_norm": 5.385923385620117,
|
| 1365 |
+
"learning_rate": 0.00019819516123077094,
|
| 1366 |
+
"loss": 5.3844,
|
| 1367 |
+
"step": 9700
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.11,
|
| 1371 |
+
"grad_norm": 4.383516788482666,
|
| 1372 |
+
"learning_rate": 0.00019816101926755305,
|
| 1373 |
+
"loss": 5.2995,
|
| 1374 |
+
"step": 9750
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.11,
|
| 1378 |
+
"grad_norm": 4.446406364440918,
|
| 1379 |
+
"learning_rate": 0.00019812585384780055,
|
| 1380 |
+
"loss": 5.386,
|
| 1381 |
+
"step": 9800
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 0.11,
|
| 1385 |
+
"grad_norm": 4.345483303070068,
|
| 1386 |
+
"learning_rate": 0.00019809035856388805,
|
| 1387 |
+
"loss": 5.2815,
|
| 1388 |
+
"step": 9850
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.11,
|
| 1392 |
+
"grad_norm": 4.791261672973633,
|
| 1393 |
+
"learning_rate": 0.00019805453353513813,
|
| 1394 |
+
"loss": 5.3757,
|
| 1395 |
+
"step": 9900
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.11,
|
| 1399 |
+
"grad_norm": 5.622151851654053,
|
| 1400 |
+
"learning_rate": 0.00019801837888198172,
|
| 1401 |
+
"loss": 5.4405,
|
| 1402 |
+
"step": 9950
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.11,
|
| 1406 |
+
"grad_norm": 4.934606075286865,
|
| 1407 |
+
"learning_rate": 0.0001979818947259579,
|
| 1408 |
+
"loss": 5.139,
|
| 1409 |
+
"step": 10000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 0.11,
|
| 1413 |
+
"grad_norm": 3.9659693241119385,
|
| 1414 |
+
"learning_rate": 0.0001979450811897134,
|
| 1415 |
+
"loss": 5.1726,
|
| 1416 |
+
"step": 10050
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 0.11,
|
| 1420 |
+
"grad_norm": 5.214992046356201,
|
| 1421 |
+
"learning_rate": 0.00019790793839700226,
|
| 1422 |
+
"loss": 5.2864,
|
| 1423 |
+
"step": 10100
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 0.11,
|
| 1427 |
+
"grad_norm": 4.5359601974487305,
|
| 1428 |
+
"learning_rate": 0.00019787046647268524,
|
| 1429 |
+
"loss": 5.1443,
|
| 1430 |
+
"step": 10150
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.11,
|
| 1434 |
+
"grad_norm": 4.26462984085083,
|
| 1435 |
+
"learning_rate": 0.00019783266554272962,
|
| 1436 |
+
"loss": 5.0597,
|
| 1437 |
+
"step": 10200
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.11,
|
| 1441 |
+
"grad_norm": 5.053945064544678,
|
| 1442 |
+
"learning_rate": 0.00019779453573420873,
|
| 1443 |
+
"loss": 5.2946,
|
| 1444 |
+
"step": 10250
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 0.11,
|
| 1448 |
+
"grad_norm": 6.082211494445801,
|
| 1449 |
+
"learning_rate": 0.00019775607717530127,
|
| 1450 |
+
"loss": 5.2075,
|
| 1451 |
+
"step": 10300
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 0.11,
|
| 1455 |
+
"grad_norm": 4.107390403747559,
|
| 1456 |
+
"learning_rate": 0.00019771728999529132,
|
| 1457 |
+
"loss": 5.1394,
|
| 1458 |
+
"step": 10350
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 0.12,
|
| 1462 |
+
"grad_norm": 4.58411169052124,
|
| 1463 |
+
"learning_rate": 0.00019767817432456752,
|
| 1464 |
+
"loss": 5.1064,
|
| 1465 |
+
"step": 10400
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.12,
|
| 1469 |
+
"grad_norm": 8.38965892791748,
|
| 1470 |
+
"learning_rate": 0.00019763952239228627,
|
| 1471 |
+
"loss": 5.0808,
|
| 1472 |
+
"step": 10450
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.12,
|
| 1476 |
+
"grad_norm": 3.885803699493408,
|
| 1477 |
+
"learning_rate": 0.00019759975669894338,
|
| 1478 |
+
"loss": 5.0664,
|
| 1479 |
+
"step": 10500
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.12,
|
| 1483 |
+
"grad_norm": 4.1605916023254395,
|
| 1484 |
+
"learning_rate": 0.00019755966290999167,
|
| 1485 |
+
"loss": 5.2469,
|
| 1486 |
+
"step": 10550
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 0.12,
|
| 1490 |
+
"grad_norm": 4.821887016296387,
|
| 1491 |
+
"learning_rate": 0.00019751924116021225,
|
| 1492 |
+
"loss": 5.2451,
|
| 1493 |
+
"step": 10600
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 0.12,
|
| 1497 |
+
"grad_norm": 3.865694761276245,
|
| 1498 |
+
"learning_rate": 0.00019747849158548858,
|
| 1499 |
+
"loss": 5.2334,
|
| 1500 |
+
"step": 10650
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 0.12,
|
| 1504 |
+
"grad_norm": 3.640681028366089,
|
| 1505 |
+
"learning_rate": 0.00019743741432280625,
|
| 1506 |
+
"loss": 5.1206,
|
| 1507 |
+
"step": 10700
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.12,
|
| 1511 |
+
"grad_norm": 4.04166316986084,
|
| 1512 |
+
"learning_rate": 0.00019739600951025236,
|
| 1513 |
+
"loss": 5.0059,
|
| 1514 |
+
"step": 10750
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.12,
|
| 1518 |
+
"grad_norm": 4.637605667114258,
|
| 1519 |
+
"learning_rate": 0.00019735427728701516,
|
| 1520 |
+
"loss": 5.0302,
|
| 1521 |
+
"step": 10800
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.12,
|
| 1525 |
+
"grad_norm": 4.08723783493042,
|
| 1526 |
+
"learning_rate": 0.0001973122177933835,
|
| 1527 |
+
"loss": 5.1551,
|
| 1528 |
+
"step": 10850
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 0.12,
|
| 1532 |
+
"grad_norm": 3.7944953441619873,
|
| 1533 |
+
"learning_rate": 0.00019726983117074643,
|
| 1534 |
+
"loss": 5.0665,
|
| 1535 |
+
"step": 10900
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 0.12,
|
| 1539 |
+
"grad_norm": 5.2847371101379395,
|
| 1540 |
+
"learning_rate": 0.00019722711756159266,
|
| 1541 |
+
"loss": 5.2212,
|
| 1542 |
+
"step": 10950
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 0.12,
|
| 1546 |
+
"grad_norm": 4.109150409698486,
|
| 1547 |
+
"learning_rate": 0.00019718407710951012,
|
| 1548 |
+
"loss": 5.2645,
|
| 1549 |
+
"step": 11000
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 0.12,
|
| 1553 |
+
"grad_norm": 4.127768039703369,
|
| 1554 |
+
"learning_rate": 0.0001971407099591855,
|
| 1555 |
+
"loss": 5.0395,
|
| 1556 |
+
"step": 11050
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.12,
|
| 1560 |
+
"grad_norm": 5.058667182922363,
|
| 1561 |
+
"learning_rate": 0.00019709701625640367,
|
| 1562 |
+
"loss": 5.0247,
|
| 1563 |
+
"step": 11100
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.12,
|
| 1567 |
+
"grad_norm": 5.4407267570495605,
|
| 1568 |
+
"learning_rate": 0.00019705299614804732,
|
| 1569 |
+
"loss": 4.9935,
|
| 1570 |
+
"step": 11150
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 0.12,
|
| 1574 |
+
"grad_norm": 3.7877707481384277,
|
| 1575 |
+
"learning_rate": 0.00019700864978209636,
|
| 1576 |
+
"loss": 5.074,
|
| 1577 |
+
"step": 11200
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 0.12,
|
| 1581 |
+
"grad_norm": 3.777330160140991,
|
| 1582 |
+
"learning_rate": 0.00019696397730762746,
|
| 1583 |
+
"loss": 5.0458,
|
| 1584 |
+
"step": 11250
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 0.13,
|
| 1588 |
+
"grad_norm": 4.143067836761475,
|
| 1589 |
+
"learning_rate": 0.0001969189788748136,
|
| 1590 |
+
"loss": 4.9375,
|
| 1591 |
+
"step": 11300
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 0.13,
|
| 1595 |
+
"grad_norm": 5.560107231140137,
|
| 1596 |
+
"learning_rate": 0.00019687365463492344,
|
| 1597 |
+
"loss": 4.8285,
|
| 1598 |
+
"step": 11350
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.13,
|
| 1602 |
+
"grad_norm": 4.057905197143555,
|
| 1603 |
+
"learning_rate": 0.00019682800474032095,
|
| 1604 |
+
"loss": 4.9753,
|
| 1605 |
+
"step": 11400
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.13,
|
| 1609 |
+
"grad_norm": 3.835442066192627,
|
| 1610 |
+
"learning_rate": 0.00019678202934446482,
|
| 1611 |
+
"loss": 4.9368,
|
| 1612 |
+
"step": 11450
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 0.13,
|
| 1616 |
+
"grad_norm": 5.135551929473877,
|
| 1617 |
+
"learning_rate": 0.0001967357286019079,
|
| 1618 |
+
"loss": 4.9994,
|
| 1619 |
+
"step": 11500
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 0.13,
|
| 1623 |
+
"grad_norm": 4.615053653717041,
|
| 1624 |
+
"learning_rate": 0.00019668910266829685,
|
| 1625 |
+
"loss": 5.0182,
|
| 1626 |
+
"step": 11550
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 0.13,
|
| 1630 |
+
"grad_norm": 4.474258899688721,
|
| 1631 |
+
"learning_rate": 0.0001966421517003714,
|
| 1632 |
+
"loss": 4.8704,
|
| 1633 |
+
"step": 11600
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 0.13,
|
| 1637 |
+
"grad_norm": 4.264945030212402,
|
| 1638 |
+
"learning_rate": 0.00019659487585596406,
|
| 1639 |
+
"loss": 4.9076,
|
| 1640 |
+
"step": 11650
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.13,
|
| 1644 |
+
"grad_norm": 4.091209411621094,
|
| 1645 |
+
"learning_rate": 0.00019654727529399925,
|
| 1646 |
+
"loss": 4.7135,
|
| 1647 |
+
"step": 11700
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.13,
|
| 1651 |
+
"grad_norm": 4.154038429260254,
|
| 1652 |
+
"learning_rate": 0.00019649935017449318,
|
| 1653 |
+
"loss": 4.8239,
|
| 1654 |
+
"step": 11750
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 0.13,
|
| 1658 |
+
"grad_norm": 3.697162628173828,
|
| 1659 |
+
"learning_rate": 0.00019645110065855305,
|
| 1660 |
+
"loss": 4.9972,
|
| 1661 |
+
"step": 11800
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 0.13,
|
| 1665 |
+
"grad_norm": 4.0024847984313965,
|
| 1666 |
+
"learning_rate": 0.00019640252690837645,
|
| 1667 |
+
"loss": 4.8854,
|
| 1668 |
+
"step": 11850
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 0.13,
|
| 1672 |
+
"grad_norm": 3.9416885375976562,
|
| 1673 |
+
"learning_rate": 0.0001963536290872511,
|
| 1674 |
+
"loss": 4.8547,
|
| 1675 |
+
"step": 11900
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 0.13,
|
| 1679 |
+
"grad_norm": 3.978651285171509,
|
| 1680 |
+
"learning_rate": 0.000196304407359554,
|
| 1681 |
+
"loss": 4.7873,
|
| 1682 |
+
"step": 11950
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 0.13,
|
| 1686 |
+
"grad_norm": 4.435175895690918,
|
| 1687 |
+
"learning_rate": 0.0001962548618907511,
|
| 1688 |
+
"loss": 4.8124,
|
| 1689 |
+
"step": 12000
|
| 1690 |
+
}
|
| 1691 |
+
],
|
| 1692 |
+
"logging_steps": 50,
|
| 1693 |
+
"max_steps": 90183,
|
| 1694 |
+
"num_input_tokens_seen": 0,
|
| 1695 |
+
"num_train_epochs": 1,
|
| 1696 |
+
"save_steps": 1000,
|
| 1697 |
+
"total_flos": 4315086323712000.0,
|
| 1698 |
+
"train_batch_size": 16,
|
| 1699 |
+
"trial_name": null,
|
| 1700 |
+
"trial_params": null
|
| 1701 |
+
}
|
checkpoints/checkpoint-12000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bcdffec292848a9e4baadecd0be02bf6760a1c3238602b1aad355cf1ba43d36e
|
| 3 |
+
size 5048
|
checkpoints/checkpoint-13000/config.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"AlbertForMaskedLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0,
|
| 6 |
+
"bos_token_id": 2,
|
| 7 |
+
"classifier_dropout_prob": 0.1,
|
| 8 |
+
"embedding_size": 128,
|
| 9 |
+
"eos_token_id": 3,
|
| 10 |
+
"hidden_act": "gelu_new",
|
| 11 |
+
"hidden_dropout_prob": 0,
|
| 12 |
+
"hidden_size": 768,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"inner_group_num": 1,
|
| 15 |
+
"intermediate_size": 3072,
|
| 16 |
+
"layer_norm_eps": 1e-12,
|
| 17 |
+
"max_position_embeddings": 512,
|
| 18 |
+
"model_type": "albert",
|
| 19 |
+
"num_attention_heads": 12,
|
| 20 |
+
"num_hidden_groups": 1,
|
| 21 |
+
"num_hidden_layers": 12,
|
| 22 |
+
"pad_token_id": 0,
|
| 23 |
+
"position_embedding_type": "absolute",
|
| 24 |
+
"torch_dtype": "float32",
|
| 25 |
+
"transformers_version": "4.38.2",
|
| 26 |
+
"type_vocab_size": 2,
|
| 27 |
+
"vocab_size": 30000
|
| 28 |
+
}
|
checkpoints/checkpoint-13000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:edc4219adab8ac3ba9103fd86a182da9a0bedb14cfe4d42bade6801c7a27a8a6
|
| 3 |
+
size 44890256
|
checkpoints/checkpoint-13000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:948e9a773fdffc69920d17092383c982e2dc52a78986bf49329517f8efbf26e0
|
| 3 |
+
size 89797322
|
checkpoints/checkpoint-13000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:19f0a906d5244d02faa3a94f1a18dfa95c5aad1b758492f4d6054a85d264b063
|
| 3 |
+
size 14244
|
checkpoints/checkpoint-13000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:75f67ff29c4600c308e522220590872f6ef4c35245be0fedfa991591dcb42d49
|
| 3 |
+
size 1064
|
checkpoints/checkpoint-13000/trainer_state.json
ADDED
|
@@ -0,0 +1,1841 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.14415133672643402,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 13000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"grad_norm": 3.2580924034118652,
|
| 14 |
+
"learning_rate": 2.2172949002217296e-06,
|
| 15 |
+
"loss": 10.2933,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.0,
|
| 20 |
+
"grad_norm": 2.4347386360168457,
|
| 21 |
+
"learning_rate": 4.434589800443459e-06,
|
| 22 |
+
"loss": 10.1894,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.0,
|
| 27 |
+
"grad_norm": 2.3895885944366455,
|
| 28 |
+
"learning_rate": 6.651884700665188e-06,
|
| 29 |
+
"loss": 10.1424,
|
| 30 |
+
"step": 150
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.0,
|
| 34 |
+
"grad_norm": 2.129647731781006,
|
| 35 |
+
"learning_rate": 8.869179600886918e-06,
|
| 36 |
+
"loss": 10.0995,
|
| 37 |
+
"step": 200
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.0,
|
| 41 |
+
"grad_norm": 2.3564186096191406,
|
| 42 |
+
"learning_rate": 1.1086474501108649e-05,
|
| 43 |
+
"loss": 10.0479,
|
| 44 |
+
"step": 250
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.0,
|
| 48 |
+
"grad_norm": 1.830551028251648,
|
| 49 |
+
"learning_rate": 1.3303769401330377e-05,
|
| 50 |
+
"loss": 9.9971,
|
| 51 |
+
"step": 300
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.0,
|
| 55 |
+
"grad_norm": 2.1173911094665527,
|
| 56 |
+
"learning_rate": 1.5521064301552106e-05,
|
| 57 |
+
"loss": 9.9201,
|
| 58 |
+
"step": 350
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.0,
|
| 62 |
+
"grad_norm": 1.6636557579040527,
|
| 63 |
+
"learning_rate": 1.7738359201773837e-05,
|
| 64 |
+
"loss": 9.8562,
|
| 65 |
+
"step": 400
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.0,
|
| 69 |
+
"grad_norm": 2.4503839015960693,
|
| 70 |
+
"learning_rate": 1.9955654101995567e-05,
|
| 71 |
+
"loss": 9.7599,
|
| 72 |
+
"step": 450
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.01,
|
| 76 |
+
"grad_norm": 1.822424054145813,
|
| 77 |
+
"learning_rate": 2.2172949002217298e-05,
|
| 78 |
+
"loss": 9.6608,
|
| 79 |
+
"step": 500
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.01,
|
| 83 |
+
"grad_norm": 1.6598998308181763,
|
| 84 |
+
"learning_rate": 2.4390243902439026e-05,
|
| 85 |
+
"loss": 9.55,
|
| 86 |
+
"step": 550
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.01,
|
| 90 |
+
"grad_norm": 1.8471707105636597,
|
| 91 |
+
"learning_rate": 2.6607538802660753e-05,
|
| 92 |
+
"loss": 9.4606,
|
| 93 |
+
"step": 600
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.01,
|
| 97 |
+
"grad_norm": 1.4833533763885498,
|
| 98 |
+
"learning_rate": 2.8824833702882487e-05,
|
| 99 |
+
"loss": 9.3283,
|
| 100 |
+
"step": 650
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.01,
|
| 104 |
+
"grad_norm": 1.688541054725647,
|
| 105 |
+
"learning_rate": 3.104212860310421e-05,
|
| 106 |
+
"loss": 9.2229,
|
| 107 |
+
"step": 700
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.01,
|
| 111 |
+
"grad_norm": 1.6466543674468994,
|
| 112 |
+
"learning_rate": 3.325942350332594e-05,
|
| 113 |
+
"loss": 9.1093,
|
| 114 |
+
"step": 750
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.01,
|
| 118 |
+
"grad_norm": 1.4169293642044067,
|
| 119 |
+
"learning_rate": 3.547671840354767e-05,
|
| 120 |
+
"loss": 8.9703,
|
| 121 |
+
"step": 800
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.01,
|
| 125 |
+
"grad_norm": 1.7079193592071533,
|
| 126 |
+
"learning_rate": 3.7694013303769404e-05,
|
| 127 |
+
"loss": 8.8351,
|
| 128 |
+
"step": 850
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.01,
|
| 132 |
+
"grad_norm": 1.5513204336166382,
|
| 133 |
+
"learning_rate": 3.9911308203991135e-05,
|
| 134 |
+
"loss": 8.7111,
|
| 135 |
+
"step": 900
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.01,
|
| 139 |
+
"grad_norm": 1.485573172569275,
|
| 140 |
+
"learning_rate": 4.212860310421286e-05,
|
| 141 |
+
"loss": 8.5627,
|
| 142 |
+
"step": 950
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.01,
|
| 146 |
+
"grad_norm": 1.511690616607666,
|
| 147 |
+
"learning_rate": 4.4345898004434597e-05,
|
| 148 |
+
"loss": 8.5042,
|
| 149 |
+
"step": 1000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.01,
|
| 153 |
+
"grad_norm": 2.1478614807128906,
|
| 154 |
+
"learning_rate": 4.656319290465632e-05,
|
| 155 |
+
"loss": 8.3287,
|
| 156 |
+
"step": 1050
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.01,
|
| 160 |
+
"grad_norm": 1.4060652256011963,
|
| 161 |
+
"learning_rate": 4.878048780487805e-05,
|
| 162 |
+
"loss": 8.2341,
|
| 163 |
+
"step": 1100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.01,
|
| 167 |
+
"grad_norm": 1.3950035572052002,
|
| 168 |
+
"learning_rate": 5.099778270509978e-05,
|
| 169 |
+
"loss": 8.1277,
|
| 170 |
+
"step": 1150
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.01,
|
| 174 |
+
"grad_norm": 1.5197688341140747,
|
| 175 |
+
"learning_rate": 5.3215077605321506e-05,
|
| 176 |
+
"loss": 8.0311,
|
| 177 |
+
"step": 1200
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.01,
|
| 181 |
+
"grad_norm": 1.3406693935394287,
|
| 182 |
+
"learning_rate": 5.543237250554324e-05,
|
| 183 |
+
"loss": 7.9824,
|
| 184 |
+
"step": 1250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.01,
|
| 188 |
+
"grad_norm": 1.4520119428634644,
|
| 189 |
+
"learning_rate": 5.7649667405764975e-05,
|
| 190 |
+
"loss": 7.9948,
|
| 191 |
+
"step": 1300
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.01,
|
| 195 |
+
"grad_norm": 1.179124116897583,
|
| 196 |
+
"learning_rate": 5.98669623059867e-05,
|
| 197 |
+
"loss": 7.9144,
|
| 198 |
+
"step": 1350
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.02,
|
| 202 |
+
"grad_norm": 1.4039533138275146,
|
| 203 |
+
"learning_rate": 6.208425720620842e-05,
|
| 204 |
+
"loss": 7.8768,
|
| 205 |
+
"step": 1400
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.02,
|
| 209 |
+
"grad_norm": 1.5542700290679932,
|
| 210 |
+
"learning_rate": 6.430155210643016e-05,
|
| 211 |
+
"loss": 7.894,
|
| 212 |
+
"step": 1450
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.02,
|
| 216 |
+
"grad_norm": 1.4150550365447998,
|
| 217 |
+
"learning_rate": 6.651884700665188e-05,
|
| 218 |
+
"loss": 7.8409,
|
| 219 |
+
"step": 1500
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.02,
|
| 223 |
+
"grad_norm": 1.6647827625274658,
|
| 224 |
+
"learning_rate": 6.873614190687362e-05,
|
| 225 |
+
"loss": 7.91,
|
| 226 |
+
"step": 1550
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.02,
|
| 230 |
+
"grad_norm": 1.7795697450637817,
|
| 231 |
+
"learning_rate": 7.095343680709535e-05,
|
| 232 |
+
"loss": 7.8256,
|
| 233 |
+
"step": 1600
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.02,
|
| 237 |
+
"grad_norm": 1.933110237121582,
|
| 238 |
+
"learning_rate": 7.317073170731707e-05,
|
| 239 |
+
"loss": 7.8463,
|
| 240 |
+
"step": 1650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.02,
|
| 244 |
+
"grad_norm": 1.1942570209503174,
|
| 245 |
+
"learning_rate": 7.538802660753881e-05,
|
| 246 |
+
"loss": 7.7827,
|
| 247 |
+
"step": 1700
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.02,
|
| 251 |
+
"grad_norm": 1.6759297847747803,
|
| 252 |
+
"learning_rate": 7.760532150776053e-05,
|
| 253 |
+
"loss": 7.8,
|
| 254 |
+
"step": 1750
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.02,
|
| 258 |
+
"grad_norm": 1.093256950378418,
|
| 259 |
+
"learning_rate": 7.982261640798227e-05,
|
| 260 |
+
"loss": 7.7461,
|
| 261 |
+
"step": 1800
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.02,
|
| 265 |
+
"grad_norm": 1.567872166633606,
|
| 266 |
+
"learning_rate": 8.2039911308204e-05,
|
| 267 |
+
"loss": 7.7338,
|
| 268 |
+
"step": 1850
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.02,
|
| 272 |
+
"grad_norm": 1.3017679452896118,
|
| 273 |
+
"learning_rate": 8.425720620842572e-05,
|
| 274 |
+
"loss": 7.804,
|
| 275 |
+
"step": 1900
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.02,
|
| 279 |
+
"grad_norm": 1.7510960102081299,
|
| 280 |
+
"learning_rate": 8.647450110864746e-05,
|
| 281 |
+
"loss": 7.7405,
|
| 282 |
+
"step": 1950
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.02,
|
| 286 |
+
"grad_norm": 1.7215120792388916,
|
| 287 |
+
"learning_rate": 8.869179600886919e-05,
|
| 288 |
+
"loss": 7.7429,
|
| 289 |
+
"step": 2000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.02,
|
| 293 |
+
"grad_norm": 1.6202715635299683,
|
| 294 |
+
"learning_rate": 9.090909090909092e-05,
|
| 295 |
+
"loss": 7.6588,
|
| 296 |
+
"step": 2050
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.02,
|
| 300 |
+
"grad_norm": 1.5680756568908691,
|
| 301 |
+
"learning_rate": 9.312638580931264e-05,
|
| 302 |
+
"loss": 7.6224,
|
| 303 |
+
"step": 2100
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.02,
|
| 307 |
+
"grad_norm": 1.462240219116211,
|
| 308 |
+
"learning_rate": 9.534368070953438e-05,
|
| 309 |
+
"loss": 7.6851,
|
| 310 |
+
"step": 2150
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.02,
|
| 314 |
+
"grad_norm": 2.2018320560455322,
|
| 315 |
+
"learning_rate": 9.75609756097561e-05,
|
| 316 |
+
"loss": 7.6443,
|
| 317 |
+
"step": 2200
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.02,
|
| 321 |
+
"grad_norm": 1.9520208835601807,
|
| 322 |
+
"learning_rate": 9.977827050997783e-05,
|
| 323 |
+
"loss": 7.6456,
|
| 324 |
+
"step": 2250
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.03,
|
| 328 |
+
"grad_norm": 1.115421175956726,
|
| 329 |
+
"learning_rate": 0.00010199556541019956,
|
| 330 |
+
"loss": 7.5894,
|
| 331 |
+
"step": 2300
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.03,
|
| 335 |
+
"grad_norm": 1.6002250909805298,
|
| 336 |
+
"learning_rate": 0.0001042128603104213,
|
| 337 |
+
"loss": 7.6017,
|
| 338 |
+
"step": 2350
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.03,
|
| 342 |
+
"grad_norm": 1.6516796350479126,
|
| 343 |
+
"learning_rate": 0.00010643015521064301,
|
| 344 |
+
"loss": 7.4548,
|
| 345 |
+
"step": 2400
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.03,
|
| 349 |
+
"grad_norm": 2.2168257236480713,
|
| 350 |
+
"learning_rate": 0.00010864745011086475,
|
| 351 |
+
"loss": 7.5867,
|
| 352 |
+
"step": 2450
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.03,
|
| 356 |
+
"grad_norm": 1.5447593927383423,
|
| 357 |
+
"learning_rate": 0.00011086474501108647,
|
| 358 |
+
"loss": 7.5317,
|
| 359 |
+
"step": 2500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.03,
|
| 363 |
+
"grad_norm": 1.6840906143188477,
|
| 364 |
+
"learning_rate": 0.00011308203991130821,
|
| 365 |
+
"loss": 7.5127,
|
| 366 |
+
"step": 2550
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.03,
|
| 370 |
+
"grad_norm": 1.2965503931045532,
|
| 371 |
+
"learning_rate": 0.00011529933481152995,
|
| 372 |
+
"loss": 7.4911,
|
| 373 |
+
"step": 2600
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.03,
|
| 377 |
+
"grad_norm": 1.643584966659546,
|
| 378 |
+
"learning_rate": 0.00011751662971175166,
|
| 379 |
+
"loss": 7.4416,
|
| 380 |
+
"step": 2650
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.03,
|
| 384 |
+
"grad_norm": 1.5419111251831055,
|
| 385 |
+
"learning_rate": 0.0001197339246119734,
|
| 386 |
+
"loss": 7.4944,
|
| 387 |
+
"step": 2700
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.03,
|
| 391 |
+
"grad_norm": 1.7774205207824707,
|
| 392 |
+
"learning_rate": 0.00012195121951219512,
|
| 393 |
+
"loss": 7.4244,
|
| 394 |
+
"step": 2750
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.03,
|
| 398 |
+
"grad_norm": 2.1709322929382324,
|
| 399 |
+
"learning_rate": 0.00012416851441241685,
|
| 400 |
+
"loss": 7.371,
|
| 401 |
+
"step": 2800
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.03,
|
| 405 |
+
"grad_norm": 1.5503411293029785,
|
| 406 |
+
"learning_rate": 0.0001263858093126386,
|
| 407 |
+
"loss": 7.3031,
|
| 408 |
+
"step": 2850
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.03,
|
| 412 |
+
"grad_norm": 1.7744035720825195,
|
| 413 |
+
"learning_rate": 0.00012860310421286032,
|
| 414 |
+
"loss": 7.3338,
|
| 415 |
+
"step": 2900
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.03,
|
| 419 |
+
"grad_norm": 2.2014000415802,
|
| 420 |
+
"learning_rate": 0.00013082039911308205,
|
| 421 |
+
"loss": 7.2962,
|
| 422 |
+
"step": 2950
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.03,
|
| 426 |
+
"grad_norm": 1.6716220378875732,
|
| 427 |
+
"learning_rate": 0.00013303769401330377,
|
| 428 |
+
"loss": 7.3348,
|
| 429 |
+
"step": 3000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.03,
|
| 433 |
+
"grad_norm": 1.7045074701309204,
|
| 434 |
+
"learning_rate": 0.0001352549889135255,
|
| 435 |
+
"loss": 7.2864,
|
| 436 |
+
"step": 3050
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.03,
|
| 440 |
+
"grad_norm": 1.8933771848678589,
|
| 441 |
+
"learning_rate": 0.00013747228381374724,
|
| 442 |
+
"loss": 7.2744,
|
| 443 |
+
"step": 3100
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.03,
|
| 447 |
+
"grad_norm": 2.298779249191284,
|
| 448 |
+
"learning_rate": 0.00013968957871396897,
|
| 449 |
+
"loss": 7.2472,
|
| 450 |
+
"step": 3150
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.04,
|
| 454 |
+
"grad_norm": 1.3420922756195068,
|
| 455 |
+
"learning_rate": 0.0001419068736141907,
|
| 456 |
+
"loss": 7.3019,
|
| 457 |
+
"step": 3200
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.04,
|
| 461 |
+
"grad_norm": 1.9339039325714111,
|
| 462 |
+
"learning_rate": 0.00014412416851441242,
|
| 463 |
+
"loss": 7.2982,
|
| 464 |
+
"step": 3250
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.04,
|
| 468 |
+
"grad_norm": 2.69667387008667,
|
| 469 |
+
"learning_rate": 0.00014634146341463414,
|
| 470 |
+
"loss": 7.2851,
|
| 471 |
+
"step": 3300
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.04,
|
| 475 |
+
"grad_norm": 2.3124189376831055,
|
| 476 |
+
"learning_rate": 0.0001485587583148559,
|
| 477 |
+
"loss": 7.258,
|
| 478 |
+
"step": 3350
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.04,
|
| 482 |
+
"grad_norm": 1.975651741027832,
|
| 483 |
+
"learning_rate": 0.00015077605321507762,
|
| 484 |
+
"loss": 7.1275,
|
| 485 |
+
"step": 3400
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.04,
|
| 489 |
+
"grad_norm": 1.9704022407531738,
|
| 490 |
+
"learning_rate": 0.00015299334811529934,
|
| 491 |
+
"loss": 7.1473,
|
| 492 |
+
"step": 3450
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.04,
|
| 496 |
+
"grad_norm": 2.5047757625579834,
|
| 497 |
+
"learning_rate": 0.00015521064301552106,
|
| 498 |
+
"loss": 7.1096,
|
| 499 |
+
"step": 3500
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.04,
|
| 503 |
+
"grad_norm": 1.5465894937515259,
|
| 504 |
+
"learning_rate": 0.0001574279379157428,
|
| 505 |
+
"loss": 7.1501,
|
| 506 |
+
"step": 3550
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.04,
|
| 510 |
+
"grad_norm": 1.9557933807373047,
|
| 511 |
+
"learning_rate": 0.00015964523281596454,
|
| 512 |
+
"loss": 7.2033,
|
| 513 |
+
"step": 3600
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.04,
|
| 517 |
+
"grad_norm": 2.420116424560547,
|
| 518 |
+
"learning_rate": 0.00016186252771618626,
|
| 519 |
+
"loss": 7.1275,
|
| 520 |
+
"step": 3650
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.04,
|
| 524 |
+
"grad_norm": 2.114737033843994,
|
| 525 |
+
"learning_rate": 0.000164079822616408,
|
| 526 |
+
"loss": 7.0932,
|
| 527 |
+
"step": 3700
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.04,
|
| 531 |
+
"grad_norm": 2.3085389137268066,
|
| 532 |
+
"learning_rate": 0.00016629711751662974,
|
| 533 |
+
"loss": 7.0311,
|
| 534 |
+
"step": 3750
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.04,
|
| 538 |
+
"grad_norm": 2.5679140090942383,
|
| 539 |
+
"learning_rate": 0.00016851441241685144,
|
| 540 |
+
"loss": 6.9168,
|
| 541 |
+
"step": 3800
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.04,
|
| 545 |
+
"grad_norm": 1.8611838817596436,
|
| 546 |
+
"learning_rate": 0.0001707317073170732,
|
| 547 |
+
"loss": 7.0085,
|
| 548 |
+
"step": 3850
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.04,
|
| 552 |
+
"grad_norm": 1.8603994846343994,
|
| 553 |
+
"learning_rate": 0.0001729490022172949,
|
| 554 |
+
"loss": 6.9432,
|
| 555 |
+
"step": 3900
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.04,
|
| 559 |
+
"grad_norm": 2.4244627952575684,
|
| 560 |
+
"learning_rate": 0.00017516629711751663,
|
| 561 |
+
"loss": 6.9333,
|
| 562 |
+
"step": 3950
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.04,
|
| 566 |
+
"grad_norm": 2.177870750427246,
|
| 567 |
+
"learning_rate": 0.00017738359201773839,
|
| 568 |
+
"loss": 6.9499,
|
| 569 |
+
"step": 4000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.04,
|
| 573 |
+
"grad_norm": 1.9320554733276367,
|
| 574 |
+
"learning_rate": 0.00017960088691796008,
|
| 575 |
+
"loss": 6.8204,
|
| 576 |
+
"step": 4050
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.05,
|
| 580 |
+
"grad_norm": 1.5062849521636963,
|
| 581 |
+
"learning_rate": 0.00018181818181818183,
|
| 582 |
+
"loss": 6.9505,
|
| 583 |
+
"step": 4100
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.05,
|
| 587 |
+
"grad_norm": 2.9272422790527344,
|
| 588 |
+
"learning_rate": 0.00018403547671840356,
|
| 589 |
+
"loss": 6.8701,
|
| 590 |
+
"step": 4150
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.05,
|
| 594 |
+
"grad_norm": 2.0309596061706543,
|
| 595 |
+
"learning_rate": 0.00018625277161862528,
|
| 596 |
+
"loss": 6.924,
|
| 597 |
+
"step": 4200
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.05,
|
| 601 |
+
"grad_norm": 2.0265886783599854,
|
| 602 |
+
"learning_rate": 0.00018847006651884703,
|
| 603 |
+
"loss": 6.9223,
|
| 604 |
+
"step": 4250
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.05,
|
| 608 |
+
"grad_norm": 2.5160486698150635,
|
| 609 |
+
"learning_rate": 0.00019068736141906876,
|
| 610 |
+
"loss": 6.8708,
|
| 611 |
+
"step": 4300
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.05,
|
| 615 |
+
"grad_norm": 2.613301992416382,
|
| 616 |
+
"learning_rate": 0.00019290465631929045,
|
| 617 |
+
"loss": 6.8937,
|
| 618 |
+
"step": 4350
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.05,
|
| 622 |
+
"grad_norm": 2.3031229972839355,
|
| 623 |
+
"learning_rate": 0.0001951219512195122,
|
| 624 |
+
"loss": 6.8337,
|
| 625 |
+
"step": 4400
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.05,
|
| 629 |
+
"grad_norm": 2.54779052734375,
|
| 630 |
+
"learning_rate": 0.00019733924611973393,
|
| 631 |
+
"loss": 6.8334,
|
| 632 |
+
"step": 4450
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.05,
|
| 636 |
+
"grad_norm": 2.8277971744537354,
|
| 637 |
+
"learning_rate": 0.00019955654101995565,
|
| 638 |
+
"loss": 6.7925,
|
| 639 |
+
"step": 4500
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.05,
|
| 643 |
+
"grad_norm": 2.0113885402679443,
|
| 644 |
+
"learning_rate": 0.00019999989242739025,
|
| 645 |
+
"loss": 6.8458,
|
| 646 |
+
"step": 4550
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.05,
|
| 650 |
+
"grad_norm": 2.2395377159118652,
|
| 651 |
+
"learning_rate": 0.00019999945541405976,
|
| 652 |
+
"loss": 6.6251,
|
| 653 |
+
"step": 4600
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.05,
|
| 657 |
+
"grad_norm": 2.445993423461914,
|
| 658 |
+
"learning_rate": 0.0001999986822381884,
|
| 659 |
+
"loss": 6.8099,
|
| 660 |
+
"step": 4650
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.05,
|
| 664 |
+
"grad_norm": 4.077752590179443,
|
| 665 |
+
"learning_rate": 0.0001999975729023753,
|
| 666 |
+
"loss": 6.8053,
|
| 667 |
+
"step": 4700
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.05,
|
| 671 |
+
"grad_norm": 3.167569875717163,
|
| 672 |
+
"learning_rate": 0.00019999612741034963,
|
| 673 |
+
"loss": 6.7706,
|
| 674 |
+
"step": 4750
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.05,
|
| 678 |
+
"grad_norm": 1.893659234046936,
|
| 679 |
+
"learning_rate": 0.00019999434576697066,
|
| 680 |
+
"loss": 6.8245,
|
| 681 |
+
"step": 4800
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.05,
|
| 685 |
+
"grad_norm": 3.6101326942443848,
|
| 686 |
+
"learning_rate": 0.00019999222797822762,
|
| 687 |
+
"loss": 6.7407,
|
| 688 |
+
"step": 4850
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.05,
|
| 692 |
+
"grad_norm": 2.2858726978302,
|
| 693 |
+
"learning_rate": 0.00019998977405123974,
|
| 694 |
+
"loss": 6.74,
|
| 695 |
+
"step": 4900
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.05,
|
| 699 |
+
"grad_norm": 1.9325459003448486,
|
| 700 |
+
"learning_rate": 0.0001999869839942563,
|
| 701 |
+
"loss": 6.716,
|
| 702 |
+
"step": 4950
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.06,
|
| 706 |
+
"grad_norm": 2.0043437480926514,
|
| 707 |
+
"learning_rate": 0.00019998385781665643,
|
| 708 |
+
"loss": 6.6003,
|
| 709 |
+
"step": 5000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.06,
|
| 713 |
+
"grad_norm": 4.151523113250732,
|
| 714 |
+
"learning_rate": 0.00019998039552894924,
|
| 715 |
+
"loss": 6.6801,
|
| 716 |
+
"step": 5050
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.06,
|
| 720 |
+
"grad_norm": 3.8407771587371826,
|
| 721 |
+
"learning_rate": 0.00019997659714277372,
|
| 722 |
+
"loss": 6.608,
|
| 723 |
+
"step": 5100
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.06,
|
| 727 |
+
"grad_norm": 2.230713129043579,
|
| 728 |
+
"learning_rate": 0.00019997246267089867,
|
| 729 |
+
"loss": 6.6479,
|
| 730 |
+
"step": 5150
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.06,
|
| 734 |
+
"grad_norm": 2.2546942234039307,
|
| 735 |
+
"learning_rate": 0.0001999679921272227,
|
| 736 |
+
"loss": 6.6548,
|
| 737 |
+
"step": 5200
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.06,
|
| 741 |
+
"grad_norm": 3.180986166000366,
|
| 742 |
+
"learning_rate": 0.00019996318552677425,
|
| 743 |
+
"loss": 6.6851,
|
| 744 |
+
"step": 5250
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.06,
|
| 748 |
+
"grad_norm": 2.341231346130371,
|
| 749 |
+
"learning_rate": 0.00019995804288571134,
|
| 750 |
+
"loss": 6.547,
|
| 751 |
+
"step": 5300
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.06,
|
| 755 |
+
"grad_norm": 3.1117124557495117,
|
| 756 |
+
"learning_rate": 0.00019995256422132172,
|
| 757 |
+
"loss": 6.7072,
|
| 758 |
+
"step": 5350
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.06,
|
| 762 |
+
"grad_norm": 2.0082530975341797,
|
| 763 |
+
"learning_rate": 0.0001999467495520227,
|
| 764 |
+
"loss": 6.5422,
|
| 765 |
+
"step": 5400
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.06,
|
| 769 |
+
"grad_norm": 2.409489870071411,
|
| 770 |
+
"learning_rate": 0.0001999405988973611,
|
| 771 |
+
"loss": 6.3716,
|
| 772 |
+
"step": 5450
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.06,
|
| 776 |
+
"grad_norm": 2.649052381515503,
|
| 777 |
+
"learning_rate": 0.00019993411227801328,
|
| 778 |
+
"loss": 6.6434,
|
| 779 |
+
"step": 5500
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.06,
|
| 783 |
+
"grad_norm": 3.081116199493408,
|
| 784 |
+
"learning_rate": 0.00019992728971578492,
|
| 785 |
+
"loss": 6.4624,
|
| 786 |
+
"step": 5550
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.06,
|
| 790 |
+
"grad_norm": 3.1578280925750732,
|
| 791 |
+
"learning_rate": 0.00019992013123361102,
|
| 792 |
+
"loss": 6.5416,
|
| 793 |
+
"step": 5600
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.06,
|
| 797 |
+
"grad_norm": 3.7874557971954346,
|
| 798 |
+
"learning_rate": 0.0001999126368555559,
|
| 799 |
+
"loss": 6.4512,
|
| 800 |
+
"step": 5650
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.06,
|
| 804 |
+
"grad_norm": 2.7693099975585938,
|
| 805 |
+
"learning_rate": 0.00019990480660681293,
|
| 806 |
+
"loss": 6.5105,
|
| 807 |
+
"step": 5700
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.06,
|
| 811 |
+
"grad_norm": 2.4338185787200928,
|
| 812 |
+
"learning_rate": 0.00019989680712666593,
|
| 813 |
+
"loss": 6.5092,
|
| 814 |
+
"step": 5750
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.06,
|
| 818 |
+
"grad_norm": 3.656937837600708,
|
| 819 |
+
"learning_rate": 0.00019988831193270577,
|
| 820 |
+
"loss": 6.4269,
|
| 821 |
+
"step": 5800
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.06,
|
| 825 |
+
"grad_norm": 2.857292652130127,
|
| 826 |
+
"learning_rate": 0.00019987948094982952,
|
| 827 |
+
"loss": 6.4387,
|
| 828 |
+
"step": 5850
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.07,
|
| 832 |
+
"grad_norm": 3.4963467121124268,
|
| 833 |
+
"learning_rate": 0.00019987031420772385,
|
| 834 |
+
"loss": 6.3851,
|
| 835 |
+
"step": 5900
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.07,
|
| 839 |
+
"grad_norm": 2.602522611618042,
|
| 840 |
+
"learning_rate": 0.00019986081173720396,
|
| 841 |
+
"loss": 6.3413,
|
| 842 |
+
"step": 5950
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.07,
|
| 846 |
+
"grad_norm": 2.6455273628234863,
|
| 847 |
+
"learning_rate": 0.00019985097357021385,
|
| 848 |
+
"loss": 6.2965,
|
| 849 |
+
"step": 6000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.07,
|
| 853 |
+
"grad_norm": 3.5592167377471924,
|
| 854 |
+
"learning_rate": 0.0001998407997398259,
|
| 855 |
+
"loss": 6.4293,
|
| 856 |
+
"step": 6050
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.07,
|
| 860 |
+
"grad_norm": 3.6016533374786377,
|
| 861 |
+
"learning_rate": 0.00019983029028024094,
|
| 862 |
+
"loss": 6.2897,
|
| 863 |
+
"step": 6100
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.07,
|
| 867 |
+
"grad_norm": 2.5536839962005615,
|
| 868 |
+
"learning_rate": 0.000199819445226788,
|
| 869 |
+
"loss": 6.3157,
|
| 870 |
+
"step": 6150
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.07,
|
| 874 |
+
"grad_norm": 2.0514349937438965,
|
| 875 |
+
"learning_rate": 0.00019980826461592427,
|
| 876 |
+
"loss": 6.3847,
|
| 877 |
+
"step": 6200
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.07,
|
| 881 |
+
"grad_norm": 2.72495174407959,
|
| 882 |
+
"learning_rate": 0.00019979674848523505,
|
| 883 |
+
"loss": 6.3517,
|
| 884 |
+
"step": 6250
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.07,
|
| 888 |
+
"grad_norm": 2.4264872074127197,
|
| 889 |
+
"learning_rate": 0.00019978489687343335,
|
| 890 |
+
"loss": 6.2533,
|
| 891 |
+
"step": 6300
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.07,
|
| 895 |
+
"grad_norm": 2.8361423015594482,
|
| 896 |
+
"learning_rate": 0.0001997727098203602,
|
| 897 |
+
"loss": 6.3654,
|
| 898 |
+
"step": 6350
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.07,
|
| 902 |
+
"grad_norm": 2.9690892696380615,
|
| 903 |
+
"learning_rate": 0.00019976018736698404,
|
| 904 |
+
"loss": 6.3968,
|
| 905 |
+
"step": 6400
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.07,
|
| 909 |
+
"grad_norm": 2.6132867336273193,
|
| 910 |
+
"learning_rate": 0.0001997473295554009,
|
| 911 |
+
"loss": 6.3444,
|
| 912 |
+
"step": 6450
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.07,
|
| 916 |
+
"grad_norm": 4.820697784423828,
|
| 917 |
+
"learning_rate": 0.00019973413642883424,
|
| 918 |
+
"loss": 6.2019,
|
| 919 |
+
"step": 6500
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.07,
|
| 923 |
+
"grad_norm": 2.2316782474517822,
|
| 924 |
+
"learning_rate": 0.00019972060803163458,
|
| 925 |
+
"loss": 6.2049,
|
| 926 |
+
"step": 6550
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.07,
|
| 930 |
+
"grad_norm": 3.9528305530548096,
|
| 931 |
+
"learning_rate": 0.00019970674440927957,
|
| 932 |
+
"loss": 6.1718,
|
| 933 |
+
"step": 6600
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.07,
|
| 937 |
+
"grad_norm": 1.891073226928711,
|
| 938 |
+
"learning_rate": 0.0001996925456083738,
|
| 939 |
+
"loss": 6.2393,
|
| 940 |
+
"step": 6650
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.07,
|
| 944 |
+
"grad_norm": 2.813270092010498,
|
| 945 |
+
"learning_rate": 0.00019967801167664853,
|
| 946 |
+
"loss": 6.2116,
|
| 947 |
+
"step": 6700
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.07,
|
| 951 |
+
"grad_norm": 2.2726826667785645,
|
| 952 |
+
"learning_rate": 0.00019966314266296173,
|
| 953 |
+
"loss": 6.1521,
|
| 954 |
+
"step": 6750
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.08,
|
| 958 |
+
"grad_norm": 2.3895318508148193,
|
| 959 |
+
"learning_rate": 0.00019964793861729772,
|
| 960 |
+
"loss": 6.1072,
|
| 961 |
+
"step": 6800
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.08,
|
| 965 |
+
"grad_norm": 3.190431833267212,
|
| 966 |
+
"learning_rate": 0.000199632399590767,
|
| 967 |
+
"loss": 6.2009,
|
| 968 |
+
"step": 6850
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.08,
|
| 972 |
+
"grad_norm": 3.79266095161438,
|
| 973 |
+
"learning_rate": 0.00019961652563560634,
|
| 974 |
+
"loss": 6.028,
|
| 975 |
+
"step": 6900
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.08,
|
| 979 |
+
"grad_norm": 3.260039806365967,
|
| 980 |
+
"learning_rate": 0.00019960031680517826,
|
| 981 |
+
"loss": 6.0733,
|
| 982 |
+
"step": 6950
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.08,
|
| 986 |
+
"grad_norm": 3.0739686489105225,
|
| 987 |
+
"learning_rate": 0.0001995837731539711,
|
| 988 |
+
"loss": 6.0521,
|
| 989 |
+
"step": 7000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.08,
|
| 993 |
+
"grad_norm": 3.0517771244049072,
|
| 994 |
+
"learning_rate": 0.00019956689473759872,
|
| 995 |
+
"loss": 6.0544,
|
| 996 |
+
"step": 7050
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.08,
|
| 1000 |
+
"grad_norm": 3.9524648189544678,
|
| 1001 |
+
"learning_rate": 0.0001995496816128003,
|
| 1002 |
+
"loss": 6.1326,
|
| 1003 |
+
"step": 7100
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.08,
|
| 1007 |
+
"grad_norm": 4.498497486114502,
|
| 1008 |
+
"learning_rate": 0.00019953213383744033,
|
| 1009 |
+
"loss": 6.236,
|
| 1010 |
+
"step": 7150
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.08,
|
| 1014 |
+
"grad_norm": 4.157576084136963,
|
| 1015 |
+
"learning_rate": 0.00019951425147050807,
|
| 1016 |
+
"loss": 5.9898,
|
| 1017 |
+
"step": 7200
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.08,
|
| 1021 |
+
"grad_norm": 3.9297516345977783,
|
| 1022 |
+
"learning_rate": 0.00019949603457211775,
|
| 1023 |
+
"loss": 6.086,
|
| 1024 |
+
"step": 7250
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.08,
|
| 1028 |
+
"grad_norm": 3.3214786052703857,
|
| 1029 |
+
"learning_rate": 0.00019947748320350804,
|
| 1030 |
+
"loss": 5.9589,
|
| 1031 |
+
"step": 7300
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.08,
|
| 1035 |
+
"grad_norm": 2.8847291469573975,
|
| 1036 |
+
"learning_rate": 0.00019945859742704201,
|
| 1037 |
+
"loss": 6.1931,
|
| 1038 |
+
"step": 7350
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.08,
|
| 1042 |
+
"grad_norm": 3.387896776199341,
|
| 1043 |
+
"learning_rate": 0.00019943937730620702,
|
| 1044 |
+
"loss": 6.0539,
|
| 1045 |
+
"step": 7400
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.08,
|
| 1049 |
+
"grad_norm": 3.1214797496795654,
|
| 1050 |
+
"learning_rate": 0.00019941982290561417,
|
| 1051 |
+
"loss": 6.0288,
|
| 1052 |
+
"step": 7450
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.08,
|
| 1056 |
+
"grad_norm": 3.7995123863220215,
|
| 1057 |
+
"learning_rate": 0.00019939993429099841,
|
| 1058 |
+
"loss": 6.0526,
|
| 1059 |
+
"step": 7500
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.08,
|
| 1063 |
+
"grad_norm": 4.788393974304199,
|
| 1064 |
+
"learning_rate": 0.00019937971152921818,
|
| 1065 |
+
"loss": 5.9799,
|
| 1066 |
+
"step": 7550
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.08,
|
| 1070 |
+
"grad_norm": 4.009220123291016,
|
| 1071 |
+
"learning_rate": 0.0001993591546882552,
|
| 1072 |
+
"loss": 6.1223,
|
| 1073 |
+
"step": 7600
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.08,
|
| 1077 |
+
"grad_norm": 3.5576276779174805,
|
| 1078 |
+
"learning_rate": 0.00019933826383721428,
|
| 1079 |
+
"loss": 5.989,
|
| 1080 |
+
"step": 7650
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.09,
|
| 1084 |
+
"grad_norm": 3.1287412643432617,
|
| 1085 |
+
"learning_rate": 0.00019931703904632294,
|
| 1086 |
+
"loss": 6.0542,
|
| 1087 |
+
"step": 7700
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.09,
|
| 1091 |
+
"grad_norm": 3.6518595218658447,
|
| 1092 |
+
"learning_rate": 0.00019929548038693146,
|
| 1093 |
+
"loss": 6.041,
|
| 1094 |
+
"step": 7750
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.09,
|
| 1098 |
+
"grad_norm": 3.268080472946167,
|
| 1099 |
+
"learning_rate": 0.0001992735879315123,
|
| 1100 |
+
"loss": 5.888,
|
| 1101 |
+
"step": 7800
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.09,
|
| 1105 |
+
"grad_norm": 3.6055593490600586,
|
| 1106 |
+
"learning_rate": 0.00019925136175366007,
|
| 1107 |
+
"loss": 5.913,
|
| 1108 |
+
"step": 7850
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.09,
|
| 1112 |
+
"grad_norm": 4.866463661193848,
|
| 1113 |
+
"learning_rate": 0.00019922880192809137,
|
| 1114 |
+
"loss": 5.9858,
|
| 1115 |
+
"step": 7900
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.09,
|
| 1119 |
+
"grad_norm": 3.44808292388916,
|
| 1120 |
+
"learning_rate": 0.00019920590853064423,
|
| 1121 |
+
"loss": 5.7686,
|
| 1122 |
+
"step": 7950
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.09,
|
| 1126 |
+
"grad_norm": 2.9507765769958496,
|
| 1127 |
+
"learning_rate": 0.00019918268163827808,
|
| 1128 |
+
"loss": 5.8557,
|
| 1129 |
+
"step": 8000
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.09,
|
| 1133 |
+
"grad_norm": 3.441870927810669,
|
| 1134 |
+
"learning_rate": 0.00019915912132907352,
|
| 1135 |
+
"loss": 5.8268,
|
| 1136 |
+
"step": 8050
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.09,
|
| 1140 |
+
"grad_norm": 3.838809013366699,
|
| 1141 |
+
"learning_rate": 0.00019913522768223182,
|
| 1142 |
+
"loss": 5.9833,
|
| 1143 |
+
"step": 8100
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.09,
|
| 1147 |
+
"grad_norm": 4.165487289428711,
|
| 1148 |
+
"learning_rate": 0.00019911100077807498,
|
| 1149 |
+
"loss": 5.7422,
|
| 1150 |
+
"step": 8150
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.09,
|
| 1154 |
+
"grad_norm": 3.5947463512420654,
|
| 1155 |
+
"learning_rate": 0.0001990864406980452,
|
| 1156 |
+
"loss": 5.7479,
|
| 1157 |
+
"step": 8200
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.09,
|
| 1161 |
+
"grad_norm": 4.130446434020996,
|
| 1162 |
+
"learning_rate": 0.00019906154752470472,
|
| 1163 |
+
"loss": 5.7767,
|
| 1164 |
+
"step": 8250
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.09,
|
| 1168 |
+
"grad_norm": 4.866550922393799,
|
| 1169 |
+
"learning_rate": 0.00019903632134173554,
|
| 1170 |
+
"loss": 5.7681,
|
| 1171 |
+
"step": 8300
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.09,
|
| 1175 |
+
"grad_norm": 3.2839725017547607,
|
| 1176 |
+
"learning_rate": 0.00019901076223393903,
|
| 1177 |
+
"loss": 5.6656,
|
| 1178 |
+
"step": 8350
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.09,
|
| 1182 |
+
"grad_norm": 3.0762476921081543,
|
| 1183 |
+
"learning_rate": 0.0001989848702872359,
|
| 1184 |
+
"loss": 5.789,
|
| 1185 |
+
"step": 8400
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.09,
|
| 1189 |
+
"grad_norm": 3.7109107971191406,
|
| 1190 |
+
"learning_rate": 0.00019895864558866556,
|
| 1191 |
+
"loss": 5.773,
|
| 1192 |
+
"step": 8450
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.09,
|
| 1196 |
+
"grad_norm": 5.400998115539551,
|
| 1197 |
+
"learning_rate": 0.00019893208822638618,
|
| 1198 |
+
"loss": 5.7506,
|
| 1199 |
+
"step": 8500
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.09,
|
| 1203 |
+
"grad_norm": 3.3062849044799805,
|
| 1204 |
+
"learning_rate": 0.00019890519828967413,
|
| 1205 |
+
"loss": 5.7515,
|
| 1206 |
+
"step": 8550
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.1,
|
| 1210 |
+
"grad_norm": 4.109920501708984,
|
| 1211 |
+
"learning_rate": 0.00019887797586892373,
|
| 1212 |
+
"loss": 5.7972,
|
| 1213 |
+
"step": 8600
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.1,
|
| 1217 |
+
"grad_norm": 3.4838390350341797,
|
| 1218 |
+
"learning_rate": 0.00019885042105564717,
|
| 1219 |
+
"loss": 5.6753,
|
| 1220 |
+
"step": 8650
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.1,
|
| 1224 |
+
"grad_norm": 4.251760959625244,
|
| 1225 |
+
"learning_rate": 0.00019882253394247381,
|
| 1226 |
+
"loss": 5.6303,
|
| 1227 |
+
"step": 8700
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.1,
|
| 1231 |
+
"grad_norm": 4.042376518249512,
|
| 1232 |
+
"learning_rate": 0.00019879431462315025,
|
| 1233 |
+
"loss": 5.5753,
|
| 1234 |
+
"step": 8750
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.1,
|
| 1238 |
+
"grad_norm": 4.239652633666992,
|
| 1239 |
+
"learning_rate": 0.0001987657631925398,
|
| 1240 |
+
"loss": 5.5335,
|
| 1241 |
+
"step": 8800
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.1,
|
| 1245 |
+
"grad_norm": 5.15481424331665,
|
| 1246 |
+
"learning_rate": 0.00019873687974662215,
|
| 1247 |
+
"loss": 5.5396,
|
| 1248 |
+
"step": 8850
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 0.1,
|
| 1252 |
+
"grad_norm": 4.36835241317749,
|
| 1253 |
+
"learning_rate": 0.00019870766438249317,
|
| 1254 |
+
"loss": 5.6017,
|
| 1255 |
+
"step": 8900
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.1,
|
| 1259 |
+
"grad_norm": 4.165258407592773,
|
| 1260 |
+
"learning_rate": 0.00019867811719836452,
|
| 1261 |
+
"loss": 5.7228,
|
| 1262 |
+
"step": 8950
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.1,
|
| 1266 |
+
"grad_norm": 4.125988006591797,
|
| 1267 |
+
"learning_rate": 0.0001986482382935633,
|
| 1268 |
+
"loss": 5.5787,
|
| 1269 |
+
"step": 9000
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.1,
|
| 1273 |
+
"grad_norm": 4.177731037139893,
|
| 1274 |
+
"learning_rate": 0.0001986180277685317,
|
| 1275 |
+
"loss": 5.5829,
|
| 1276 |
+
"step": 9050
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.1,
|
| 1280 |
+
"grad_norm": 5.006561279296875,
|
| 1281 |
+
"learning_rate": 0.00019858748572482683,
|
| 1282 |
+
"loss": 5.5466,
|
| 1283 |
+
"step": 9100
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.1,
|
| 1287 |
+
"grad_norm": 4.33070182800293,
|
| 1288 |
+
"learning_rate": 0.00019855661226512007,
|
| 1289 |
+
"loss": 5.5544,
|
| 1290 |
+
"step": 9150
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.1,
|
| 1294 |
+
"grad_norm": 4.358560085296631,
|
| 1295 |
+
"learning_rate": 0.00019852540749319708,
|
| 1296 |
+
"loss": 5.4599,
|
| 1297 |
+
"step": 9200
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 0.1,
|
| 1301 |
+
"grad_norm": 4.536096096038818,
|
| 1302 |
+
"learning_rate": 0.00019849387151395708,
|
| 1303 |
+
"loss": 5.4983,
|
| 1304 |
+
"step": 9250
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.1,
|
| 1308 |
+
"grad_norm": 4.66163444519043,
|
| 1309 |
+
"learning_rate": 0.0001984620044334129,
|
| 1310 |
+
"loss": 5.4097,
|
| 1311 |
+
"step": 9300
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.1,
|
| 1315 |
+
"grad_norm": 4.4319233894348145,
|
| 1316 |
+
"learning_rate": 0.00019842980635869024,
|
| 1317 |
+
"loss": 5.4093,
|
| 1318 |
+
"step": 9350
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 0.1,
|
| 1322 |
+
"grad_norm": 4.98419713973999,
|
| 1323 |
+
"learning_rate": 0.0001983972773980276,
|
| 1324 |
+
"loss": 5.4056,
|
| 1325 |
+
"step": 9400
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.1,
|
| 1329 |
+
"grad_norm": 3.6354339122772217,
|
| 1330 |
+
"learning_rate": 0.0001983644176607757,
|
| 1331 |
+
"loss": 5.3171,
|
| 1332 |
+
"step": 9450
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 0.11,
|
| 1336 |
+
"grad_norm": 4.495342254638672,
|
| 1337 |
+
"learning_rate": 0.00019833122725739736,
|
| 1338 |
+
"loss": 5.4521,
|
| 1339 |
+
"step": 9500
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 0.11,
|
| 1343 |
+
"grad_norm": 4.5558671951293945,
|
| 1344 |
+
"learning_rate": 0.00019829770629946678,
|
| 1345 |
+
"loss": 5.5158,
|
| 1346 |
+
"step": 9550
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.11,
|
| 1350 |
+
"grad_norm": 3.7165732383728027,
|
| 1351 |
+
"learning_rate": 0.00019826385489966957,
|
| 1352 |
+
"loss": 5.301,
|
| 1353 |
+
"step": 9600
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.11,
|
| 1357 |
+
"grad_norm": 6.030915260314941,
|
| 1358 |
+
"learning_rate": 0.00019822967317180204,
|
| 1359 |
+
"loss": 5.3316,
|
| 1360 |
+
"step": 9650
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.11,
|
| 1364 |
+
"grad_norm": 5.385923385620117,
|
| 1365 |
+
"learning_rate": 0.00019819516123077094,
|
| 1366 |
+
"loss": 5.3844,
|
| 1367 |
+
"step": 9700
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.11,
|
| 1371 |
+
"grad_norm": 4.383516788482666,
|
| 1372 |
+
"learning_rate": 0.00019816101926755305,
|
| 1373 |
+
"loss": 5.2995,
|
| 1374 |
+
"step": 9750
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.11,
|
| 1378 |
+
"grad_norm": 4.446406364440918,
|
| 1379 |
+
"learning_rate": 0.00019812585384780055,
|
| 1380 |
+
"loss": 5.386,
|
| 1381 |
+
"step": 9800
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 0.11,
|
| 1385 |
+
"grad_norm": 4.345483303070068,
|
| 1386 |
+
"learning_rate": 0.00019809035856388805,
|
| 1387 |
+
"loss": 5.2815,
|
| 1388 |
+
"step": 9850
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.11,
|
| 1392 |
+
"grad_norm": 4.791261672973633,
|
| 1393 |
+
"learning_rate": 0.00019805453353513813,
|
| 1394 |
+
"loss": 5.3757,
|
| 1395 |
+
"step": 9900
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.11,
|
| 1399 |
+
"grad_norm": 5.622151851654053,
|
| 1400 |
+
"learning_rate": 0.00019801837888198172,
|
| 1401 |
+
"loss": 5.4405,
|
| 1402 |
+
"step": 9950
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.11,
|
| 1406 |
+
"grad_norm": 4.934606075286865,
|
| 1407 |
+
"learning_rate": 0.0001979818947259579,
|
| 1408 |
+
"loss": 5.139,
|
| 1409 |
+
"step": 10000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 0.11,
|
| 1413 |
+
"grad_norm": 3.9659693241119385,
|
| 1414 |
+
"learning_rate": 0.0001979450811897134,
|
| 1415 |
+
"loss": 5.1726,
|
| 1416 |
+
"step": 10050
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 0.11,
|
| 1420 |
+
"grad_norm": 5.214992046356201,
|
| 1421 |
+
"learning_rate": 0.00019790793839700226,
|
| 1422 |
+
"loss": 5.2864,
|
| 1423 |
+
"step": 10100
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 0.11,
|
| 1427 |
+
"grad_norm": 4.5359601974487305,
|
| 1428 |
+
"learning_rate": 0.00019787046647268524,
|
| 1429 |
+
"loss": 5.1443,
|
| 1430 |
+
"step": 10150
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.11,
|
| 1434 |
+
"grad_norm": 4.26462984085083,
|
| 1435 |
+
"learning_rate": 0.00019783266554272962,
|
| 1436 |
+
"loss": 5.0597,
|
| 1437 |
+
"step": 10200
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.11,
|
| 1441 |
+
"grad_norm": 5.053945064544678,
|
| 1442 |
+
"learning_rate": 0.00019779453573420873,
|
| 1443 |
+
"loss": 5.2946,
|
| 1444 |
+
"step": 10250
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 0.11,
|
| 1448 |
+
"grad_norm": 6.082211494445801,
|
| 1449 |
+
"learning_rate": 0.00019775607717530127,
|
| 1450 |
+
"loss": 5.2075,
|
| 1451 |
+
"step": 10300
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 0.11,
|
| 1455 |
+
"grad_norm": 4.107390403747559,
|
| 1456 |
+
"learning_rate": 0.00019771728999529132,
|
| 1457 |
+
"loss": 5.1394,
|
| 1458 |
+
"step": 10350
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 0.12,
|
| 1462 |
+
"grad_norm": 4.58411169052124,
|
| 1463 |
+
"learning_rate": 0.00019767817432456752,
|
| 1464 |
+
"loss": 5.1064,
|
| 1465 |
+
"step": 10400
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.12,
|
| 1469 |
+
"grad_norm": 8.38965892791748,
|
| 1470 |
+
"learning_rate": 0.00019763952239228627,
|
| 1471 |
+
"loss": 5.0808,
|
| 1472 |
+
"step": 10450
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.12,
|
| 1476 |
+
"grad_norm": 3.885803699493408,
|
| 1477 |
+
"learning_rate": 0.00019759975669894338,
|
| 1478 |
+
"loss": 5.0664,
|
| 1479 |
+
"step": 10500
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.12,
|
| 1483 |
+
"grad_norm": 4.1605916023254395,
|
| 1484 |
+
"learning_rate": 0.00019755966290999167,
|
| 1485 |
+
"loss": 5.2469,
|
| 1486 |
+
"step": 10550
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 0.12,
|
| 1490 |
+
"grad_norm": 4.821887016296387,
|
| 1491 |
+
"learning_rate": 0.00019751924116021225,
|
| 1492 |
+
"loss": 5.2451,
|
| 1493 |
+
"step": 10600
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 0.12,
|
| 1497 |
+
"grad_norm": 3.865694761276245,
|
| 1498 |
+
"learning_rate": 0.00019747849158548858,
|
| 1499 |
+
"loss": 5.2334,
|
| 1500 |
+
"step": 10650
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 0.12,
|
| 1504 |
+
"grad_norm": 3.640681028366089,
|
| 1505 |
+
"learning_rate": 0.00019743741432280625,
|
| 1506 |
+
"loss": 5.1206,
|
| 1507 |
+
"step": 10700
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.12,
|
| 1511 |
+
"grad_norm": 4.04166316986084,
|
| 1512 |
+
"learning_rate": 0.00019739600951025236,
|
| 1513 |
+
"loss": 5.0059,
|
| 1514 |
+
"step": 10750
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.12,
|
| 1518 |
+
"grad_norm": 4.637605667114258,
|
| 1519 |
+
"learning_rate": 0.00019735427728701516,
|
| 1520 |
+
"loss": 5.0302,
|
| 1521 |
+
"step": 10800
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.12,
|
| 1525 |
+
"grad_norm": 4.08723783493042,
|
| 1526 |
+
"learning_rate": 0.0001973122177933835,
|
| 1527 |
+
"loss": 5.1551,
|
| 1528 |
+
"step": 10850
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 0.12,
|
| 1532 |
+
"grad_norm": 3.7944953441619873,
|
| 1533 |
+
"learning_rate": 0.00019726983117074643,
|
| 1534 |
+
"loss": 5.0665,
|
| 1535 |
+
"step": 10900
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 0.12,
|
| 1539 |
+
"grad_norm": 5.2847371101379395,
|
| 1540 |
+
"learning_rate": 0.00019722711756159266,
|
| 1541 |
+
"loss": 5.2212,
|
| 1542 |
+
"step": 10950
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 0.12,
|
| 1546 |
+
"grad_norm": 4.109150409698486,
|
| 1547 |
+
"learning_rate": 0.00019718407710951012,
|
| 1548 |
+
"loss": 5.2645,
|
| 1549 |
+
"step": 11000
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 0.12,
|
| 1553 |
+
"grad_norm": 4.127768039703369,
|
| 1554 |
+
"learning_rate": 0.0001971407099591855,
|
| 1555 |
+
"loss": 5.0395,
|
| 1556 |
+
"step": 11050
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.12,
|
| 1560 |
+
"grad_norm": 5.058667182922363,
|
| 1561 |
+
"learning_rate": 0.00019709701625640367,
|
| 1562 |
+
"loss": 5.0247,
|
| 1563 |
+
"step": 11100
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.12,
|
| 1567 |
+
"grad_norm": 5.4407267570495605,
|
| 1568 |
+
"learning_rate": 0.00019705299614804732,
|
| 1569 |
+
"loss": 4.9935,
|
| 1570 |
+
"step": 11150
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 0.12,
|
| 1574 |
+
"grad_norm": 3.7877707481384277,
|
| 1575 |
+
"learning_rate": 0.00019700864978209636,
|
| 1576 |
+
"loss": 5.074,
|
| 1577 |
+
"step": 11200
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 0.12,
|
| 1581 |
+
"grad_norm": 3.777330160140991,
|
| 1582 |
+
"learning_rate": 0.00019696397730762746,
|
| 1583 |
+
"loss": 5.0458,
|
| 1584 |
+
"step": 11250
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 0.13,
|
| 1588 |
+
"grad_norm": 4.143067836761475,
|
| 1589 |
+
"learning_rate": 0.0001969189788748136,
|
| 1590 |
+
"loss": 4.9375,
|
| 1591 |
+
"step": 11300
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 0.13,
|
| 1595 |
+
"grad_norm": 5.560107231140137,
|
| 1596 |
+
"learning_rate": 0.00019687365463492344,
|
| 1597 |
+
"loss": 4.8285,
|
| 1598 |
+
"step": 11350
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.13,
|
| 1602 |
+
"grad_norm": 4.057905197143555,
|
| 1603 |
+
"learning_rate": 0.00019682800474032095,
|
| 1604 |
+
"loss": 4.9753,
|
| 1605 |
+
"step": 11400
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.13,
|
| 1609 |
+
"grad_norm": 3.835442066192627,
|
| 1610 |
+
"learning_rate": 0.00019678202934446482,
|
| 1611 |
+
"loss": 4.9368,
|
| 1612 |
+
"step": 11450
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 0.13,
|
| 1616 |
+
"grad_norm": 5.135551929473877,
|
| 1617 |
+
"learning_rate": 0.0001967357286019079,
|
| 1618 |
+
"loss": 4.9994,
|
| 1619 |
+
"step": 11500
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 0.13,
|
| 1623 |
+
"grad_norm": 4.615053653717041,
|
| 1624 |
+
"learning_rate": 0.00019668910266829685,
|
| 1625 |
+
"loss": 5.0182,
|
| 1626 |
+
"step": 11550
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 0.13,
|
| 1630 |
+
"grad_norm": 4.474258899688721,
|
| 1631 |
+
"learning_rate": 0.0001966421517003714,
|
| 1632 |
+
"loss": 4.8704,
|
| 1633 |
+
"step": 11600
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 0.13,
|
| 1637 |
+
"grad_norm": 4.264945030212402,
|
| 1638 |
+
"learning_rate": 0.00019659487585596406,
|
| 1639 |
+
"loss": 4.9076,
|
| 1640 |
+
"step": 11650
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.13,
|
| 1644 |
+
"grad_norm": 4.091209411621094,
|
| 1645 |
+
"learning_rate": 0.00019654727529399925,
|
| 1646 |
+
"loss": 4.7135,
|
| 1647 |
+
"step": 11700
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.13,
|
| 1651 |
+
"grad_norm": 4.154038429260254,
|
| 1652 |
+
"learning_rate": 0.00019649935017449318,
|
| 1653 |
+
"loss": 4.8239,
|
| 1654 |
+
"step": 11750
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 0.13,
|
| 1658 |
+
"grad_norm": 3.697162628173828,
|
| 1659 |
+
"learning_rate": 0.00019645110065855305,
|
| 1660 |
+
"loss": 4.9972,
|
| 1661 |
+
"step": 11800
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 0.13,
|
| 1665 |
+
"grad_norm": 4.0024847984313965,
|
| 1666 |
+
"learning_rate": 0.00019640252690837645,
|
| 1667 |
+
"loss": 4.8854,
|
| 1668 |
+
"step": 11850
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 0.13,
|
| 1672 |
+
"grad_norm": 3.9416885375976562,
|
| 1673 |
+
"learning_rate": 0.0001963536290872511,
|
| 1674 |
+
"loss": 4.8547,
|
| 1675 |
+
"step": 11900
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 0.13,
|
| 1679 |
+
"grad_norm": 3.978651285171509,
|
| 1680 |
+
"learning_rate": 0.000196304407359554,
|
| 1681 |
+
"loss": 4.7873,
|
| 1682 |
+
"step": 11950
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 0.13,
|
| 1686 |
+
"grad_norm": 4.435175895690918,
|
| 1687 |
+
"learning_rate": 0.0001962548618907511,
|
| 1688 |
+
"loss": 4.8124,
|
| 1689 |
+
"step": 12000
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 0.13,
|
| 1693 |
+
"grad_norm": 3.8776824474334717,
|
| 1694 |
+
"learning_rate": 0.00019620499284739662,
|
| 1695 |
+
"loss": 4.8896,
|
| 1696 |
+
"step": 12050
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 0.13,
|
| 1700 |
+
"grad_norm": 5.041496276855469,
|
| 1701 |
+
"learning_rate": 0.00019615480039713248,
|
| 1702 |
+
"loss": 4.8343,
|
| 1703 |
+
"step": 12100
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 0.13,
|
| 1707 |
+
"grad_norm": 4.18281888961792,
|
| 1708 |
+
"learning_rate": 0.00019610428470868784,
|
| 1709 |
+
"loss": 4.8559,
|
| 1710 |
+
"step": 12150
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 0.14,
|
| 1714 |
+
"grad_norm": 4.223630905151367,
|
| 1715 |
+
"learning_rate": 0.00019605344595187844,
|
| 1716 |
+
"loss": 4.8153,
|
| 1717 |
+
"step": 12200
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 0.14,
|
| 1721 |
+
"grad_norm": 4.63677453994751,
|
| 1722 |
+
"learning_rate": 0.0001960022842976061,
|
| 1723 |
+
"loss": 4.7951,
|
| 1724 |
+
"step": 12250
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 0.14,
|
| 1728 |
+
"grad_norm": 4.188296794891357,
|
| 1729 |
+
"learning_rate": 0.00019595079991785802,
|
| 1730 |
+
"loss": 4.8904,
|
| 1731 |
+
"step": 12300
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 0.14,
|
| 1735 |
+
"grad_norm": 4.402559280395508,
|
| 1736 |
+
"learning_rate": 0.00019589899298570634,
|
| 1737 |
+
"loss": 4.7851,
|
| 1738 |
+
"step": 12350
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.14,
|
| 1742 |
+
"grad_norm": 5.976877212524414,
|
| 1743 |
+
"learning_rate": 0.00019584686367530755,
|
| 1744 |
+
"loss": 4.6431,
|
| 1745 |
+
"step": 12400
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 0.14,
|
| 1749 |
+
"grad_norm": 4.849298477172852,
|
| 1750 |
+
"learning_rate": 0.0001957944121619018,
|
| 1751 |
+
"loss": 4.7544,
|
| 1752 |
+
"step": 12450
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 0.14,
|
| 1756 |
+
"grad_norm": 4.932714462280273,
|
| 1757 |
+
"learning_rate": 0.0001957416386218124,
|
| 1758 |
+
"loss": 4.6811,
|
| 1759 |
+
"step": 12500
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 0.14,
|
| 1763 |
+
"grad_norm": 4.682474136352539,
|
| 1764 |
+
"learning_rate": 0.00019568854323244515,
|
| 1765 |
+
"loss": 4.799,
|
| 1766 |
+
"step": 12550
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 0.14,
|
| 1770 |
+
"grad_norm": 5.228520393371582,
|
| 1771 |
+
"learning_rate": 0.00019563619766470511,
|
| 1772 |
+
"loss": 4.7622,
|
| 1773 |
+
"step": 12600
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 0.14,
|
| 1777 |
+
"grad_norm": 4.093870162963867,
|
| 1778 |
+
"learning_rate": 0.00019558246554138458,
|
| 1779 |
+
"loss": 4.7369,
|
| 1780 |
+
"step": 12650
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 0.14,
|
| 1784 |
+
"grad_norm": 5.248356342315674,
|
| 1785 |
+
"learning_rate": 0.0001955284121038694,
|
| 1786 |
+
"loss": 4.7519,
|
| 1787 |
+
"step": 12700
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 0.14,
|
| 1791 |
+
"grad_norm": 3.924299955368042,
|
| 1792 |
+
"learning_rate": 0.00019547403753386803,
|
| 1793 |
+
"loss": 4.6441,
|
| 1794 |
+
"step": 12750
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 0.14,
|
| 1798 |
+
"grad_norm": 4.972569942474365,
|
| 1799 |
+
"learning_rate": 0.00019542043906868188,
|
| 1800 |
+
"loss": 4.7192,
|
| 1801 |
+
"step": 12800
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 0.14,
|
| 1805 |
+
"grad_norm": 5.033604145050049,
|
| 1806 |
+
"learning_rate": 0.00019536542919665846,
|
| 1807 |
+
"loss": 4.6397,
|
| 1808 |
+
"step": 12850
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 0.14,
|
| 1812 |
+
"grad_norm": 5.222695350646973,
|
| 1813 |
+
"learning_rate": 0.00019531009874003928,
|
| 1814 |
+
"loss": 4.6309,
|
| 1815 |
+
"step": 12900
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 0.14,
|
| 1819 |
+
"grad_norm": 3.810999631881714,
|
| 1820 |
+
"learning_rate": 0.00019525444788482562,
|
| 1821 |
+
"loss": 4.6513,
|
| 1822 |
+
"step": 12950
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 0.14,
|
| 1826 |
+
"grad_norm": 5.272600173950195,
|
| 1827 |
+
"learning_rate": 0.00019519847681809585,
|
| 1828 |
+
"loss": 4.8001,
|
| 1829 |
+
"step": 13000
|
| 1830 |
+
}
|
| 1831 |
+
],
|
| 1832 |
+
"logging_steps": 50,
|
| 1833 |
+
"max_steps": 90183,
|
| 1834 |
+
"num_input_tokens_seen": 0,
|
| 1835 |
+
"num_train_epochs": 1,
|
| 1836 |
+
"save_steps": 1000,
|
| 1837 |
+
"total_flos": 4674676850688000.0,
|
| 1838 |
+
"train_batch_size": 16,
|
| 1839 |
+
"trial_name": null,
|
| 1840 |
+
"trial_params": null
|
| 1841 |
+
}
|
checkpoints/checkpoint-13000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bcdffec292848a9e4baadecd0be02bf6760a1c3238602b1aad355cf1ba43d36e
|
| 3 |
+
size 5048
|
checkpoints/checkpoint-14000/config.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"AlbertForMaskedLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0,
|
| 6 |
+
"bos_token_id": 2,
|
| 7 |
+
"classifier_dropout_prob": 0.1,
|
| 8 |
+
"embedding_size": 128,
|
| 9 |
+
"eos_token_id": 3,
|
| 10 |
+
"hidden_act": "gelu_new",
|
| 11 |
+
"hidden_dropout_prob": 0,
|
| 12 |
+
"hidden_size": 768,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"inner_group_num": 1,
|
| 15 |
+
"intermediate_size": 3072,
|
| 16 |
+
"layer_norm_eps": 1e-12,
|
| 17 |
+
"max_position_embeddings": 512,
|
| 18 |
+
"model_type": "albert",
|
| 19 |
+
"num_attention_heads": 12,
|
| 20 |
+
"num_hidden_groups": 1,
|
| 21 |
+
"num_hidden_layers": 12,
|
| 22 |
+
"pad_token_id": 0,
|
| 23 |
+
"position_embedding_type": "absolute",
|
| 24 |
+
"torch_dtype": "float32",
|
| 25 |
+
"transformers_version": "4.38.2",
|
| 26 |
+
"type_vocab_size": 2,
|
| 27 |
+
"vocab_size": 30000
|
| 28 |
+
}
|
checkpoints/checkpoint-14000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5e0a3efb18a87263970f630bf8170263f502b54358886171fe7e3c20a2f85321
|
| 3 |
+
size 44890256
|
checkpoints/checkpoint-14000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cb99c266aee02fa5c4688c5926ae8a2ce5a1f962b7b2c3bfa9b19635f228854e
|
| 3 |
+
size 89797322
|
checkpoints/checkpoint-14000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5deb8d284dfe93bb109954d8d0cbafa495adcd95bd2e73da5489dec584e1412f
|
| 3 |
+
size 14244
|
checkpoints/checkpoint-14000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c68ba22e1abd8d688618570a3b335f443e848ce4cbbf2603bc3fd29edb7042fa
|
| 3 |
+
size 1064
|
checkpoints/checkpoint-14000/trainer_state.json
ADDED
|
@@ -0,0 +1,1981 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.15523990109000588,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 14000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"grad_norm": 3.2580924034118652,
|
| 14 |
+
"learning_rate": 2.2172949002217296e-06,
|
| 15 |
+
"loss": 10.2933,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.0,
|
| 20 |
+
"grad_norm": 2.4347386360168457,
|
| 21 |
+
"learning_rate": 4.434589800443459e-06,
|
| 22 |
+
"loss": 10.1894,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.0,
|
| 27 |
+
"grad_norm": 2.3895885944366455,
|
| 28 |
+
"learning_rate": 6.651884700665188e-06,
|
| 29 |
+
"loss": 10.1424,
|
| 30 |
+
"step": 150
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.0,
|
| 34 |
+
"grad_norm": 2.129647731781006,
|
| 35 |
+
"learning_rate": 8.869179600886918e-06,
|
| 36 |
+
"loss": 10.0995,
|
| 37 |
+
"step": 200
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.0,
|
| 41 |
+
"grad_norm": 2.3564186096191406,
|
| 42 |
+
"learning_rate": 1.1086474501108649e-05,
|
| 43 |
+
"loss": 10.0479,
|
| 44 |
+
"step": 250
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.0,
|
| 48 |
+
"grad_norm": 1.830551028251648,
|
| 49 |
+
"learning_rate": 1.3303769401330377e-05,
|
| 50 |
+
"loss": 9.9971,
|
| 51 |
+
"step": 300
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.0,
|
| 55 |
+
"grad_norm": 2.1173911094665527,
|
| 56 |
+
"learning_rate": 1.5521064301552106e-05,
|
| 57 |
+
"loss": 9.9201,
|
| 58 |
+
"step": 350
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.0,
|
| 62 |
+
"grad_norm": 1.6636557579040527,
|
| 63 |
+
"learning_rate": 1.7738359201773837e-05,
|
| 64 |
+
"loss": 9.8562,
|
| 65 |
+
"step": 400
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.0,
|
| 69 |
+
"grad_norm": 2.4503839015960693,
|
| 70 |
+
"learning_rate": 1.9955654101995567e-05,
|
| 71 |
+
"loss": 9.7599,
|
| 72 |
+
"step": 450
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.01,
|
| 76 |
+
"grad_norm": 1.822424054145813,
|
| 77 |
+
"learning_rate": 2.2172949002217298e-05,
|
| 78 |
+
"loss": 9.6608,
|
| 79 |
+
"step": 500
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.01,
|
| 83 |
+
"grad_norm": 1.6598998308181763,
|
| 84 |
+
"learning_rate": 2.4390243902439026e-05,
|
| 85 |
+
"loss": 9.55,
|
| 86 |
+
"step": 550
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.01,
|
| 90 |
+
"grad_norm": 1.8471707105636597,
|
| 91 |
+
"learning_rate": 2.6607538802660753e-05,
|
| 92 |
+
"loss": 9.4606,
|
| 93 |
+
"step": 600
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.01,
|
| 97 |
+
"grad_norm": 1.4833533763885498,
|
| 98 |
+
"learning_rate": 2.8824833702882487e-05,
|
| 99 |
+
"loss": 9.3283,
|
| 100 |
+
"step": 650
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.01,
|
| 104 |
+
"grad_norm": 1.688541054725647,
|
| 105 |
+
"learning_rate": 3.104212860310421e-05,
|
| 106 |
+
"loss": 9.2229,
|
| 107 |
+
"step": 700
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.01,
|
| 111 |
+
"grad_norm": 1.6466543674468994,
|
| 112 |
+
"learning_rate": 3.325942350332594e-05,
|
| 113 |
+
"loss": 9.1093,
|
| 114 |
+
"step": 750
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.01,
|
| 118 |
+
"grad_norm": 1.4169293642044067,
|
| 119 |
+
"learning_rate": 3.547671840354767e-05,
|
| 120 |
+
"loss": 8.9703,
|
| 121 |
+
"step": 800
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.01,
|
| 125 |
+
"grad_norm": 1.7079193592071533,
|
| 126 |
+
"learning_rate": 3.7694013303769404e-05,
|
| 127 |
+
"loss": 8.8351,
|
| 128 |
+
"step": 850
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.01,
|
| 132 |
+
"grad_norm": 1.5513204336166382,
|
| 133 |
+
"learning_rate": 3.9911308203991135e-05,
|
| 134 |
+
"loss": 8.7111,
|
| 135 |
+
"step": 900
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.01,
|
| 139 |
+
"grad_norm": 1.485573172569275,
|
| 140 |
+
"learning_rate": 4.212860310421286e-05,
|
| 141 |
+
"loss": 8.5627,
|
| 142 |
+
"step": 950
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.01,
|
| 146 |
+
"grad_norm": 1.511690616607666,
|
| 147 |
+
"learning_rate": 4.4345898004434597e-05,
|
| 148 |
+
"loss": 8.5042,
|
| 149 |
+
"step": 1000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.01,
|
| 153 |
+
"grad_norm": 2.1478614807128906,
|
| 154 |
+
"learning_rate": 4.656319290465632e-05,
|
| 155 |
+
"loss": 8.3287,
|
| 156 |
+
"step": 1050
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.01,
|
| 160 |
+
"grad_norm": 1.4060652256011963,
|
| 161 |
+
"learning_rate": 4.878048780487805e-05,
|
| 162 |
+
"loss": 8.2341,
|
| 163 |
+
"step": 1100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.01,
|
| 167 |
+
"grad_norm": 1.3950035572052002,
|
| 168 |
+
"learning_rate": 5.099778270509978e-05,
|
| 169 |
+
"loss": 8.1277,
|
| 170 |
+
"step": 1150
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.01,
|
| 174 |
+
"grad_norm": 1.5197688341140747,
|
| 175 |
+
"learning_rate": 5.3215077605321506e-05,
|
| 176 |
+
"loss": 8.0311,
|
| 177 |
+
"step": 1200
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.01,
|
| 181 |
+
"grad_norm": 1.3406693935394287,
|
| 182 |
+
"learning_rate": 5.543237250554324e-05,
|
| 183 |
+
"loss": 7.9824,
|
| 184 |
+
"step": 1250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.01,
|
| 188 |
+
"grad_norm": 1.4520119428634644,
|
| 189 |
+
"learning_rate": 5.7649667405764975e-05,
|
| 190 |
+
"loss": 7.9948,
|
| 191 |
+
"step": 1300
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.01,
|
| 195 |
+
"grad_norm": 1.179124116897583,
|
| 196 |
+
"learning_rate": 5.98669623059867e-05,
|
| 197 |
+
"loss": 7.9144,
|
| 198 |
+
"step": 1350
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.02,
|
| 202 |
+
"grad_norm": 1.4039533138275146,
|
| 203 |
+
"learning_rate": 6.208425720620842e-05,
|
| 204 |
+
"loss": 7.8768,
|
| 205 |
+
"step": 1400
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.02,
|
| 209 |
+
"grad_norm": 1.5542700290679932,
|
| 210 |
+
"learning_rate": 6.430155210643016e-05,
|
| 211 |
+
"loss": 7.894,
|
| 212 |
+
"step": 1450
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.02,
|
| 216 |
+
"grad_norm": 1.4150550365447998,
|
| 217 |
+
"learning_rate": 6.651884700665188e-05,
|
| 218 |
+
"loss": 7.8409,
|
| 219 |
+
"step": 1500
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.02,
|
| 223 |
+
"grad_norm": 1.6647827625274658,
|
| 224 |
+
"learning_rate": 6.873614190687362e-05,
|
| 225 |
+
"loss": 7.91,
|
| 226 |
+
"step": 1550
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.02,
|
| 230 |
+
"grad_norm": 1.7795697450637817,
|
| 231 |
+
"learning_rate": 7.095343680709535e-05,
|
| 232 |
+
"loss": 7.8256,
|
| 233 |
+
"step": 1600
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.02,
|
| 237 |
+
"grad_norm": 1.933110237121582,
|
| 238 |
+
"learning_rate": 7.317073170731707e-05,
|
| 239 |
+
"loss": 7.8463,
|
| 240 |
+
"step": 1650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.02,
|
| 244 |
+
"grad_norm": 1.1942570209503174,
|
| 245 |
+
"learning_rate": 7.538802660753881e-05,
|
| 246 |
+
"loss": 7.7827,
|
| 247 |
+
"step": 1700
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.02,
|
| 251 |
+
"grad_norm": 1.6759297847747803,
|
| 252 |
+
"learning_rate": 7.760532150776053e-05,
|
| 253 |
+
"loss": 7.8,
|
| 254 |
+
"step": 1750
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.02,
|
| 258 |
+
"grad_norm": 1.093256950378418,
|
| 259 |
+
"learning_rate": 7.982261640798227e-05,
|
| 260 |
+
"loss": 7.7461,
|
| 261 |
+
"step": 1800
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.02,
|
| 265 |
+
"grad_norm": 1.567872166633606,
|
| 266 |
+
"learning_rate": 8.2039911308204e-05,
|
| 267 |
+
"loss": 7.7338,
|
| 268 |
+
"step": 1850
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.02,
|
| 272 |
+
"grad_norm": 1.3017679452896118,
|
| 273 |
+
"learning_rate": 8.425720620842572e-05,
|
| 274 |
+
"loss": 7.804,
|
| 275 |
+
"step": 1900
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.02,
|
| 279 |
+
"grad_norm": 1.7510960102081299,
|
| 280 |
+
"learning_rate": 8.647450110864746e-05,
|
| 281 |
+
"loss": 7.7405,
|
| 282 |
+
"step": 1950
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.02,
|
| 286 |
+
"grad_norm": 1.7215120792388916,
|
| 287 |
+
"learning_rate": 8.869179600886919e-05,
|
| 288 |
+
"loss": 7.7429,
|
| 289 |
+
"step": 2000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.02,
|
| 293 |
+
"grad_norm": 1.6202715635299683,
|
| 294 |
+
"learning_rate": 9.090909090909092e-05,
|
| 295 |
+
"loss": 7.6588,
|
| 296 |
+
"step": 2050
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.02,
|
| 300 |
+
"grad_norm": 1.5680756568908691,
|
| 301 |
+
"learning_rate": 9.312638580931264e-05,
|
| 302 |
+
"loss": 7.6224,
|
| 303 |
+
"step": 2100
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.02,
|
| 307 |
+
"grad_norm": 1.462240219116211,
|
| 308 |
+
"learning_rate": 9.534368070953438e-05,
|
| 309 |
+
"loss": 7.6851,
|
| 310 |
+
"step": 2150
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.02,
|
| 314 |
+
"grad_norm": 2.2018320560455322,
|
| 315 |
+
"learning_rate": 9.75609756097561e-05,
|
| 316 |
+
"loss": 7.6443,
|
| 317 |
+
"step": 2200
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.02,
|
| 321 |
+
"grad_norm": 1.9520208835601807,
|
| 322 |
+
"learning_rate": 9.977827050997783e-05,
|
| 323 |
+
"loss": 7.6456,
|
| 324 |
+
"step": 2250
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.03,
|
| 328 |
+
"grad_norm": 1.115421175956726,
|
| 329 |
+
"learning_rate": 0.00010199556541019956,
|
| 330 |
+
"loss": 7.5894,
|
| 331 |
+
"step": 2300
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.03,
|
| 335 |
+
"grad_norm": 1.6002250909805298,
|
| 336 |
+
"learning_rate": 0.0001042128603104213,
|
| 337 |
+
"loss": 7.6017,
|
| 338 |
+
"step": 2350
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.03,
|
| 342 |
+
"grad_norm": 1.6516796350479126,
|
| 343 |
+
"learning_rate": 0.00010643015521064301,
|
| 344 |
+
"loss": 7.4548,
|
| 345 |
+
"step": 2400
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.03,
|
| 349 |
+
"grad_norm": 2.2168257236480713,
|
| 350 |
+
"learning_rate": 0.00010864745011086475,
|
| 351 |
+
"loss": 7.5867,
|
| 352 |
+
"step": 2450
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.03,
|
| 356 |
+
"grad_norm": 1.5447593927383423,
|
| 357 |
+
"learning_rate": 0.00011086474501108647,
|
| 358 |
+
"loss": 7.5317,
|
| 359 |
+
"step": 2500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.03,
|
| 363 |
+
"grad_norm": 1.6840906143188477,
|
| 364 |
+
"learning_rate": 0.00011308203991130821,
|
| 365 |
+
"loss": 7.5127,
|
| 366 |
+
"step": 2550
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.03,
|
| 370 |
+
"grad_norm": 1.2965503931045532,
|
| 371 |
+
"learning_rate": 0.00011529933481152995,
|
| 372 |
+
"loss": 7.4911,
|
| 373 |
+
"step": 2600
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.03,
|
| 377 |
+
"grad_norm": 1.643584966659546,
|
| 378 |
+
"learning_rate": 0.00011751662971175166,
|
| 379 |
+
"loss": 7.4416,
|
| 380 |
+
"step": 2650
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.03,
|
| 384 |
+
"grad_norm": 1.5419111251831055,
|
| 385 |
+
"learning_rate": 0.0001197339246119734,
|
| 386 |
+
"loss": 7.4944,
|
| 387 |
+
"step": 2700
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.03,
|
| 391 |
+
"grad_norm": 1.7774205207824707,
|
| 392 |
+
"learning_rate": 0.00012195121951219512,
|
| 393 |
+
"loss": 7.4244,
|
| 394 |
+
"step": 2750
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.03,
|
| 398 |
+
"grad_norm": 2.1709322929382324,
|
| 399 |
+
"learning_rate": 0.00012416851441241685,
|
| 400 |
+
"loss": 7.371,
|
| 401 |
+
"step": 2800
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.03,
|
| 405 |
+
"grad_norm": 1.5503411293029785,
|
| 406 |
+
"learning_rate": 0.0001263858093126386,
|
| 407 |
+
"loss": 7.3031,
|
| 408 |
+
"step": 2850
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.03,
|
| 412 |
+
"grad_norm": 1.7744035720825195,
|
| 413 |
+
"learning_rate": 0.00012860310421286032,
|
| 414 |
+
"loss": 7.3338,
|
| 415 |
+
"step": 2900
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.03,
|
| 419 |
+
"grad_norm": 2.2014000415802,
|
| 420 |
+
"learning_rate": 0.00013082039911308205,
|
| 421 |
+
"loss": 7.2962,
|
| 422 |
+
"step": 2950
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.03,
|
| 426 |
+
"grad_norm": 1.6716220378875732,
|
| 427 |
+
"learning_rate": 0.00013303769401330377,
|
| 428 |
+
"loss": 7.3348,
|
| 429 |
+
"step": 3000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.03,
|
| 433 |
+
"grad_norm": 1.7045074701309204,
|
| 434 |
+
"learning_rate": 0.0001352549889135255,
|
| 435 |
+
"loss": 7.2864,
|
| 436 |
+
"step": 3050
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.03,
|
| 440 |
+
"grad_norm": 1.8933771848678589,
|
| 441 |
+
"learning_rate": 0.00013747228381374724,
|
| 442 |
+
"loss": 7.2744,
|
| 443 |
+
"step": 3100
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.03,
|
| 447 |
+
"grad_norm": 2.298779249191284,
|
| 448 |
+
"learning_rate": 0.00013968957871396897,
|
| 449 |
+
"loss": 7.2472,
|
| 450 |
+
"step": 3150
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.04,
|
| 454 |
+
"grad_norm": 1.3420922756195068,
|
| 455 |
+
"learning_rate": 0.0001419068736141907,
|
| 456 |
+
"loss": 7.3019,
|
| 457 |
+
"step": 3200
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.04,
|
| 461 |
+
"grad_norm": 1.9339039325714111,
|
| 462 |
+
"learning_rate": 0.00014412416851441242,
|
| 463 |
+
"loss": 7.2982,
|
| 464 |
+
"step": 3250
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.04,
|
| 468 |
+
"grad_norm": 2.69667387008667,
|
| 469 |
+
"learning_rate": 0.00014634146341463414,
|
| 470 |
+
"loss": 7.2851,
|
| 471 |
+
"step": 3300
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.04,
|
| 475 |
+
"grad_norm": 2.3124189376831055,
|
| 476 |
+
"learning_rate": 0.0001485587583148559,
|
| 477 |
+
"loss": 7.258,
|
| 478 |
+
"step": 3350
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.04,
|
| 482 |
+
"grad_norm": 1.975651741027832,
|
| 483 |
+
"learning_rate": 0.00015077605321507762,
|
| 484 |
+
"loss": 7.1275,
|
| 485 |
+
"step": 3400
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.04,
|
| 489 |
+
"grad_norm": 1.9704022407531738,
|
| 490 |
+
"learning_rate": 0.00015299334811529934,
|
| 491 |
+
"loss": 7.1473,
|
| 492 |
+
"step": 3450
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.04,
|
| 496 |
+
"grad_norm": 2.5047757625579834,
|
| 497 |
+
"learning_rate": 0.00015521064301552106,
|
| 498 |
+
"loss": 7.1096,
|
| 499 |
+
"step": 3500
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.04,
|
| 503 |
+
"grad_norm": 1.5465894937515259,
|
| 504 |
+
"learning_rate": 0.0001574279379157428,
|
| 505 |
+
"loss": 7.1501,
|
| 506 |
+
"step": 3550
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.04,
|
| 510 |
+
"grad_norm": 1.9557933807373047,
|
| 511 |
+
"learning_rate": 0.00015964523281596454,
|
| 512 |
+
"loss": 7.2033,
|
| 513 |
+
"step": 3600
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.04,
|
| 517 |
+
"grad_norm": 2.420116424560547,
|
| 518 |
+
"learning_rate": 0.00016186252771618626,
|
| 519 |
+
"loss": 7.1275,
|
| 520 |
+
"step": 3650
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.04,
|
| 524 |
+
"grad_norm": 2.114737033843994,
|
| 525 |
+
"learning_rate": 0.000164079822616408,
|
| 526 |
+
"loss": 7.0932,
|
| 527 |
+
"step": 3700
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.04,
|
| 531 |
+
"grad_norm": 2.3085389137268066,
|
| 532 |
+
"learning_rate": 0.00016629711751662974,
|
| 533 |
+
"loss": 7.0311,
|
| 534 |
+
"step": 3750
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.04,
|
| 538 |
+
"grad_norm": 2.5679140090942383,
|
| 539 |
+
"learning_rate": 0.00016851441241685144,
|
| 540 |
+
"loss": 6.9168,
|
| 541 |
+
"step": 3800
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.04,
|
| 545 |
+
"grad_norm": 1.8611838817596436,
|
| 546 |
+
"learning_rate": 0.0001707317073170732,
|
| 547 |
+
"loss": 7.0085,
|
| 548 |
+
"step": 3850
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.04,
|
| 552 |
+
"grad_norm": 1.8603994846343994,
|
| 553 |
+
"learning_rate": 0.0001729490022172949,
|
| 554 |
+
"loss": 6.9432,
|
| 555 |
+
"step": 3900
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.04,
|
| 559 |
+
"grad_norm": 2.4244627952575684,
|
| 560 |
+
"learning_rate": 0.00017516629711751663,
|
| 561 |
+
"loss": 6.9333,
|
| 562 |
+
"step": 3950
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.04,
|
| 566 |
+
"grad_norm": 2.177870750427246,
|
| 567 |
+
"learning_rate": 0.00017738359201773839,
|
| 568 |
+
"loss": 6.9499,
|
| 569 |
+
"step": 4000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.04,
|
| 573 |
+
"grad_norm": 1.9320554733276367,
|
| 574 |
+
"learning_rate": 0.00017960088691796008,
|
| 575 |
+
"loss": 6.8204,
|
| 576 |
+
"step": 4050
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.05,
|
| 580 |
+
"grad_norm": 1.5062849521636963,
|
| 581 |
+
"learning_rate": 0.00018181818181818183,
|
| 582 |
+
"loss": 6.9505,
|
| 583 |
+
"step": 4100
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.05,
|
| 587 |
+
"grad_norm": 2.9272422790527344,
|
| 588 |
+
"learning_rate": 0.00018403547671840356,
|
| 589 |
+
"loss": 6.8701,
|
| 590 |
+
"step": 4150
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.05,
|
| 594 |
+
"grad_norm": 2.0309596061706543,
|
| 595 |
+
"learning_rate": 0.00018625277161862528,
|
| 596 |
+
"loss": 6.924,
|
| 597 |
+
"step": 4200
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.05,
|
| 601 |
+
"grad_norm": 2.0265886783599854,
|
| 602 |
+
"learning_rate": 0.00018847006651884703,
|
| 603 |
+
"loss": 6.9223,
|
| 604 |
+
"step": 4250
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.05,
|
| 608 |
+
"grad_norm": 2.5160486698150635,
|
| 609 |
+
"learning_rate": 0.00019068736141906876,
|
| 610 |
+
"loss": 6.8708,
|
| 611 |
+
"step": 4300
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.05,
|
| 615 |
+
"grad_norm": 2.613301992416382,
|
| 616 |
+
"learning_rate": 0.00019290465631929045,
|
| 617 |
+
"loss": 6.8937,
|
| 618 |
+
"step": 4350
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.05,
|
| 622 |
+
"grad_norm": 2.3031229972839355,
|
| 623 |
+
"learning_rate": 0.0001951219512195122,
|
| 624 |
+
"loss": 6.8337,
|
| 625 |
+
"step": 4400
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.05,
|
| 629 |
+
"grad_norm": 2.54779052734375,
|
| 630 |
+
"learning_rate": 0.00019733924611973393,
|
| 631 |
+
"loss": 6.8334,
|
| 632 |
+
"step": 4450
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.05,
|
| 636 |
+
"grad_norm": 2.8277971744537354,
|
| 637 |
+
"learning_rate": 0.00019955654101995565,
|
| 638 |
+
"loss": 6.7925,
|
| 639 |
+
"step": 4500
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.05,
|
| 643 |
+
"grad_norm": 2.0113885402679443,
|
| 644 |
+
"learning_rate": 0.00019999989242739025,
|
| 645 |
+
"loss": 6.8458,
|
| 646 |
+
"step": 4550
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.05,
|
| 650 |
+
"grad_norm": 2.2395377159118652,
|
| 651 |
+
"learning_rate": 0.00019999945541405976,
|
| 652 |
+
"loss": 6.6251,
|
| 653 |
+
"step": 4600
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.05,
|
| 657 |
+
"grad_norm": 2.445993423461914,
|
| 658 |
+
"learning_rate": 0.0001999986822381884,
|
| 659 |
+
"loss": 6.8099,
|
| 660 |
+
"step": 4650
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.05,
|
| 664 |
+
"grad_norm": 4.077752590179443,
|
| 665 |
+
"learning_rate": 0.0001999975729023753,
|
| 666 |
+
"loss": 6.8053,
|
| 667 |
+
"step": 4700
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.05,
|
| 671 |
+
"grad_norm": 3.167569875717163,
|
| 672 |
+
"learning_rate": 0.00019999612741034963,
|
| 673 |
+
"loss": 6.7706,
|
| 674 |
+
"step": 4750
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.05,
|
| 678 |
+
"grad_norm": 1.893659234046936,
|
| 679 |
+
"learning_rate": 0.00019999434576697066,
|
| 680 |
+
"loss": 6.8245,
|
| 681 |
+
"step": 4800
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.05,
|
| 685 |
+
"grad_norm": 3.6101326942443848,
|
| 686 |
+
"learning_rate": 0.00019999222797822762,
|
| 687 |
+
"loss": 6.7407,
|
| 688 |
+
"step": 4850
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.05,
|
| 692 |
+
"grad_norm": 2.2858726978302,
|
| 693 |
+
"learning_rate": 0.00019998977405123974,
|
| 694 |
+
"loss": 6.74,
|
| 695 |
+
"step": 4900
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.05,
|
| 699 |
+
"grad_norm": 1.9325459003448486,
|
| 700 |
+
"learning_rate": 0.0001999869839942563,
|
| 701 |
+
"loss": 6.716,
|
| 702 |
+
"step": 4950
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.06,
|
| 706 |
+
"grad_norm": 2.0043437480926514,
|
| 707 |
+
"learning_rate": 0.00019998385781665643,
|
| 708 |
+
"loss": 6.6003,
|
| 709 |
+
"step": 5000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.06,
|
| 713 |
+
"grad_norm": 4.151523113250732,
|
| 714 |
+
"learning_rate": 0.00019998039552894924,
|
| 715 |
+
"loss": 6.6801,
|
| 716 |
+
"step": 5050
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.06,
|
| 720 |
+
"grad_norm": 3.8407771587371826,
|
| 721 |
+
"learning_rate": 0.00019997659714277372,
|
| 722 |
+
"loss": 6.608,
|
| 723 |
+
"step": 5100
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.06,
|
| 727 |
+
"grad_norm": 2.230713129043579,
|
| 728 |
+
"learning_rate": 0.00019997246267089867,
|
| 729 |
+
"loss": 6.6479,
|
| 730 |
+
"step": 5150
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.06,
|
| 734 |
+
"grad_norm": 2.2546942234039307,
|
| 735 |
+
"learning_rate": 0.0001999679921272227,
|
| 736 |
+
"loss": 6.6548,
|
| 737 |
+
"step": 5200
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.06,
|
| 741 |
+
"grad_norm": 3.180986166000366,
|
| 742 |
+
"learning_rate": 0.00019996318552677425,
|
| 743 |
+
"loss": 6.6851,
|
| 744 |
+
"step": 5250
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.06,
|
| 748 |
+
"grad_norm": 2.341231346130371,
|
| 749 |
+
"learning_rate": 0.00019995804288571134,
|
| 750 |
+
"loss": 6.547,
|
| 751 |
+
"step": 5300
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.06,
|
| 755 |
+
"grad_norm": 3.1117124557495117,
|
| 756 |
+
"learning_rate": 0.00019995256422132172,
|
| 757 |
+
"loss": 6.7072,
|
| 758 |
+
"step": 5350
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.06,
|
| 762 |
+
"grad_norm": 2.0082530975341797,
|
| 763 |
+
"learning_rate": 0.0001999467495520227,
|
| 764 |
+
"loss": 6.5422,
|
| 765 |
+
"step": 5400
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.06,
|
| 769 |
+
"grad_norm": 2.409489870071411,
|
| 770 |
+
"learning_rate": 0.0001999405988973611,
|
| 771 |
+
"loss": 6.3716,
|
| 772 |
+
"step": 5450
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.06,
|
| 776 |
+
"grad_norm": 2.649052381515503,
|
| 777 |
+
"learning_rate": 0.00019993411227801328,
|
| 778 |
+
"loss": 6.6434,
|
| 779 |
+
"step": 5500
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.06,
|
| 783 |
+
"grad_norm": 3.081116199493408,
|
| 784 |
+
"learning_rate": 0.00019992728971578492,
|
| 785 |
+
"loss": 6.4624,
|
| 786 |
+
"step": 5550
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.06,
|
| 790 |
+
"grad_norm": 3.1578280925750732,
|
| 791 |
+
"learning_rate": 0.00019992013123361102,
|
| 792 |
+
"loss": 6.5416,
|
| 793 |
+
"step": 5600
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.06,
|
| 797 |
+
"grad_norm": 3.7874557971954346,
|
| 798 |
+
"learning_rate": 0.0001999126368555559,
|
| 799 |
+
"loss": 6.4512,
|
| 800 |
+
"step": 5650
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.06,
|
| 804 |
+
"grad_norm": 2.7693099975585938,
|
| 805 |
+
"learning_rate": 0.00019990480660681293,
|
| 806 |
+
"loss": 6.5105,
|
| 807 |
+
"step": 5700
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.06,
|
| 811 |
+
"grad_norm": 2.4338185787200928,
|
| 812 |
+
"learning_rate": 0.00019989680712666593,
|
| 813 |
+
"loss": 6.5092,
|
| 814 |
+
"step": 5750
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.06,
|
| 818 |
+
"grad_norm": 3.656937837600708,
|
| 819 |
+
"learning_rate": 0.00019988831193270577,
|
| 820 |
+
"loss": 6.4269,
|
| 821 |
+
"step": 5800
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.06,
|
| 825 |
+
"grad_norm": 2.857292652130127,
|
| 826 |
+
"learning_rate": 0.00019987948094982952,
|
| 827 |
+
"loss": 6.4387,
|
| 828 |
+
"step": 5850
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.07,
|
| 832 |
+
"grad_norm": 3.4963467121124268,
|
| 833 |
+
"learning_rate": 0.00019987031420772385,
|
| 834 |
+
"loss": 6.3851,
|
| 835 |
+
"step": 5900
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.07,
|
| 839 |
+
"grad_norm": 2.602522611618042,
|
| 840 |
+
"learning_rate": 0.00019986081173720396,
|
| 841 |
+
"loss": 6.3413,
|
| 842 |
+
"step": 5950
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.07,
|
| 846 |
+
"grad_norm": 2.6455273628234863,
|
| 847 |
+
"learning_rate": 0.00019985097357021385,
|
| 848 |
+
"loss": 6.2965,
|
| 849 |
+
"step": 6000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.07,
|
| 853 |
+
"grad_norm": 3.5592167377471924,
|
| 854 |
+
"learning_rate": 0.0001998407997398259,
|
| 855 |
+
"loss": 6.4293,
|
| 856 |
+
"step": 6050
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.07,
|
| 860 |
+
"grad_norm": 3.6016533374786377,
|
| 861 |
+
"learning_rate": 0.00019983029028024094,
|
| 862 |
+
"loss": 6.2897,
|
| 863 |
+
"step": 6100
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.07,
|
| 867 |
+
"grad_norm": 2.5536839962005615,
|
| 868 |
+
"learning_rate": 0.000199819445226788,
|
| 869 |
+
"loss": 6.3157,
|
| 870 |
+
"step": 6150
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.07,
|
| 874 |
+
"grad_norm": 2.0514349937438965,
|
| 875 |
+
"learning_rate": 0.00019980826461592427,
|
| 876 |
+
"loss": 6.3847,
|
| 877 |
+
"step": 6200
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.07,
|
| 881 |
+
"grad_norm": 2.72495174407959,
|
| 882 |
+
"learning_rate": 0.00019979674848523505,
|
| 883 |
+
"loss": 6.3517,
|
| 884 |
+
"step": 6250
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.07,
|
| 888 |
+
"grad_norm": 2.4264872074127197,
|
| 889 |
+
"learning_rate": 0.00019978489687343335,
|
| 890 |
+
"loss": 6.2533,
|
| 891 |
+
"step": 6300
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.07,
|
| 895 |
+
"grad_norm": 2.8361423015594482,
|
| 896 |
+
"learning_rate": 0.0001997727098203602,
|
| 897 |
+
"loss": 6.3654,
|
| 898 |
+
"step": 6350
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.07,
|
| 902 |
+
"grad_norm": 2.9690892696380615,
|
| 903 |
+
"learning_rate": 0.00019976018736698404,
|
| 904 |
+
"loss": 6.3968,
|
| 905 |
+
"step": 6400
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.07,
|
| 909 |
+
"grad_norm": 2.6132867336273193,
|
| 910 |
+
"learning_rate": 0.0001997473295554009,
|
| 911 |
+
"loss": 6.3444,
|
| 912 |
+
"step": 6450
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.07,
|
| 916 |
+
"grad_norm": 4.820697784423828,
|
| 917 |
+
"learning_rate": 0.00019973413642883424,
|
| 918 |
+
"loss": 6.2019,
|
| 919 |
+
"step": 6500
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.07,
|
| 923 |
+
"grad_norm": 2.2316782474517822,
|
| 924 |
+
"learning_rate": 0.00019972060803163458,
|
| 925 |
+
"loss": 6.2049,
|
| 926 |
+
"step": 6550
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.07,
|
| 930 |
+
"grad_norm": 3.9528305530548096,
|
| 931 |
+
"learning_rate": 0.00019970674440927957,
|
| 932 |
+
"loss": 6.1718,
|
| 933 |
+
"step": 6600
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.07,
|
| 937 |
+
"grad_norm": 1.891073226928711,
|
| 938 |
+
"learning_rate": 0.0001996925456083738,
|
| 939 |
+
"loss": 6.2393,
|
| 940 |
+
"step": 6650
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.07,
|
| 944 |
+
"grad_norm": 2.813270092010498,
|
| 945 |
+
"learning_rate": 0.00019967801167664853,
|
| 946 |
+
"loss": 6.2116,
|
| 947 |
+
"step": 6700
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.07,
|
| 951 |
+
"grad_norm": 2.2726826667785645,
|
| 952 |
+
"learning_rate": 0.00019966314266296173,
|
| 953 |
+
"loss": 6.1521,
|
| 954 |
+
"step": 6750
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.08,
|
| 958 |
+
"grad_norm": 2.3895318508148193,
|
| 959 |
+
"learning_rate": 0.00019964793861729772,
|
| 960 |
+
"loss": 6.1072,
|
| 961 |
+
"step": 6800
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.08,
|
| 965 |
+
"grad_norm": 3.190431833267212,
|
| 966 |
+
"learning_rate": 0.000199632399590767,
|
| 967 |
+
"loss": 6.2009,
|
| 968 |
+
"step": 6850
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.08,
|
| 972 |
+
"grad_norm": 3.79266095161438,
|
| 973 |
+
"learning_rate": 0.00019961652563560634,
|
| 974 |
+
"loss": 6.028,
|
| 975 |
+
"step": 6900
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.08,
|
| 979 |
+
"grad_norm": 3.260039806365967,
|
| 980 |
+
"learning_rate": 0.00019960031680517826,
|
| 981 |
+
"loss": 6.0733,
|
| 982 |
+
"step": 6950
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.08,
|
| 986 |
+
"grad_norm": 3.0739686489105225,
|
| 987 |
+
"learning_rate": 0.0001995837731539711,
|
| 988 |
+
"loss": 6.0521,
|
| 989 |
+
"step": 7000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.08,
|
| 993 |
+
"grad_norm": 3.0517771244049072,
|
| 994 |
+
"learning_rate": 0.00019956689473759872,
|
| 995 |
+
"loss": 6.0544,
|
| 996 |
+
"step": 7050
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.08,
|
| 1000 |
+
"grad_norm": 3.9524648189544678,
|
| 1001 |
+
"learning_rate": 0.0001995496816128003,
|
| 1002 |
+
"loss": 6.1326,
|
| 1003 |
+
"step": 7100
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.08,
|
| 1007 |
+
"grad_norm": 4.498497486114502,
|
| 1008 |
+
"learning_rate": 0.00019953213383744033,
|
| 1009 |
+
"loss": 6.236,
|
| 1010 |
+
"step": 7150
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.08,
|
| 1014 |
+
"grad_norm": 4.157576084136963,
|
| 1015 |
+
"learning_rate": 0.00019951425147050807,
|
| 1016 |
+
"loss": 5.9898,
|
| 1017 |
+
"step": 7200
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.08,
|
| 1021 |
+
"grad_norm": 3.9297516345977783,
|
| 1022 |
+
"learning_rate": 0.00019949603457211775,
|
| 1023 |
+
"loss": 6.086,
|
| 1024 |
+
"step": 7250
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.08,
|
| 1028 |
+
"grad_norm": 3.3214786052703857,
|
| 1029 |
+
"learning_rate": 0.00019947748320350804,
|
| 1030 |
+
"loss": 5.9589,
|
| 1031 |
+
"step": 7300
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.08,
|
| 1035 |
+
"grad_norm": 2.8847291469573975,
|
| 1036 |
+
"learning_rate": 0.00019945859742704201,
|
| 1037 |
+
"loss": 6.1931,
|
| 1038 |
+
"step": 7350
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.08,
|
| 1042 |
+
"grad_norm": 3.387896776199341,
|
| 1043 |
+
"learning_rate": 0.00019943937730620702,
|
| 1044 |
+
"loss": 6.0539,
|
| 1045 |
+
"step": 7400
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.08,
|
| 1049 |
+
"grad_norm": 3.1214797496795654,
|
| 1050 |
+
"learning_rate": 0.00019941982290561417,
|
| 1051 |
+
"loss": 6.0288,
|
| 1052 |
+
"step": 7450
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.08,
|
| 1056 |
+
"grad_norm": 3.7995123863220215,
|
| 1057 |
+
"learning_rate": 0.00019939993429099841,
|
| 1058 |
+
"loss": 6.0526,
|
| 1059 |
+
"step": 7500
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.08,
|
| 1063 |
+
"grad_norm": 4.788393974304199,
|
| 1064 |
+
"learning_rate": 0.00019937971152921818,
|
| 1065 |
+
"loss": 5.9799,
|
| 1066 |
+
"step": 7550
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.08,
|
| 1070 |
+
"grad_norm": 4.009220123291016,
|
| 1071 |
+
"learning_rate": 0.0001993591546882552,
|
| 1072 |
+
"loss": 6.1223,
|
| 1073 |
+
"step": 7600
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.08,
|
| 1077 |
+
"grad_norm": 3.5576276779174805,
|
| 1078 |
+
"learning_rate": 0.00019933826383721428,
|
| 1079 |
+
"loss": 5.989,
|
| 1080 |
+
"step": 7650
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.09,
|
| 1084 |
+
"grad_norm": 3.1287412643432617,
|
| 1085 |
+
"learning_rate": 0.00019931703904632294,
|
| 1086 |
+
"loss": 6.0542,
|
| 1087 |
+
"step": 7700
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.09,
|
| 1091 |
+
"grad_norm": 3.6518595218658447,
|
| 1092 |
+
"learning_rate": 0.00019929548038693146,
|
| 1093 |
+
"loss": 6.041,
|
| 1094 |
+
"step": 7750
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.09,
|
| 1098 |
+
"grad_norm": 3.268080472946167,
|
| 1099 |
+
"learning_rate": 0.0001992735879315123,
|
| 1100 |
+
"loss": 5.888,
|
| 1101 |
+
"step": 7800
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.09,
|
| 1105 |
+
"grad_norm": 3.6055593490600586,
|
| 1106 |
+
"learning_rate": 0.00019925136175366007,
|
| 1107 |
+
"loss": 5.913,
|
| 1108 |
+
"step": 7850
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.09,
|
| 1112 |
+
"grad_norm": 4.866463661193848,
|
| 1113 |
+
"learning_rate": 0.00019922880192809137,
|
| 1114 |
+
"loss": 5.9858,
|
| 1115 |
+
"step": 7900
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.09,
|
| 1119 |
+
"grad_norm": 3.44808292388916,
|
| 1120 |
+
"learning_rate": 0.00019920590853064423,
|
| 1121 |
+
"loss": 5.7686,
|
| 1122 |
+
"step": 7950
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.09,
|
| 1126 |
+
"grad_norm": 2.9507765769958496,
|
| 1127 |
+
"learning_rate": 0.00019918268163827808,
|
| 1128 |
+
"loss": 5.8557,
|
| 1129 |
+
"step": 8000
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.09,
|
| 1133 |
+
"grad_norm": 3.441870927810669,
|
| 1134 |
+
"learning_rate": 0.00019915912132907352,
|
| 1135 |
+
"loss": 5.8268,
|
| 1136 |
+
"step": 8050
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.09,
|
| 1140 |
+
"grad_norm": 3.838809013366699,
|
| 1141 |
+
"learning_rate": 0.00019913522768223182,
|
| 1142 |
+
"loss": 5.9833,
|
| 1143 |
+
"step": 8100
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.09,
|
| 1147 |
+
"grad_norm": 4.165487289428711,
|
| 1148 |
+
"learning_rate": 0.00019911100077807498,
|
| 1149 |
+
"loss": 5.7422,
|
| 1150 |
+
"step": 8150
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.09,
|
| 1154 |
+
"grad_norm": 3.5947463512420654,
|
| 1155 |
+
"learning_rate": 0.0001990864406980452,
|
| 1156 |
+
"loss": 5.7479,
|
| 1157 |
+
"step": 8200
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.09,
|
| 1161 |
+
"grad_norm": 4.130446434020996,
|
| 1162 |
+
"learning_rate": 0.00019906154752470472,
|
| 1163 |
+
"loss": 5.7767,
|
| 1164 |
+
"step": 8250
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.09,
|
| 1168 |
+
"grad_norm": 4.866550922393799,
|
| 1169 |
+
"learning_rate": 0.00019903632134173554,
|
| 1170 |
+
"loss": 5.7681,
|
| 1171 |
+
"step": 8300
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.09,
|
| 1175 |
+
"grad_norm": 3.2839725017547607,
|
| 1176 |
+
"learning_rate": 0.00019901076223393903,
|
| 1177 |
+
"loss": 5.6656,
|
| 1178 |
+
"step": 8350
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.09,
|
| 1182 |
+
"grad_norm": 3.0762476921081543,
|
| 1183 |
+
"learning_rate": 0.0001989848702872359,
|
| 1184 |
+
"loss": 5.789,
|
| 1185 |
+
"step": 8400
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.09,
|
| 1189 |
+
"grad_norm": 3.7109107971191406,
|
| 1190 |
+
"learning_rate": 0.00019895864558866556,
|
| 1191 |
+
"loss": 5.773,
|
| 1192 |
+
"step": 8450
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.09,
|
| 1196 |
+
"grad_norm": 5.400998115539551,
|
| 1197 |
+
"learning_rate": 0.00019893208822638618,
|
| 1198 |
+
"loss": 5.7506,
|
| 1199 |
+
"step": 8500
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.09,
|
| 1203 |
+
"grad_norm": 3.3062849044799805,
|
| 1204 |
+
"learning_rate": 0.00019890519828967413,
|
| 1205 |
+
"loss": 5.7515,
|
| 1206 |
+
"step": 8550
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.1,
|
| 1210 |
+
"grad_norm": 4.109920501708984,
|
| 1211 |
+
"learning_rate": 0.00019887797586892373,
|
| 1212 |
+
"loss": 5.7972,
|
| 1213 |
+
"step": 8600
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.1,
|
| 1217 |
+
"grad_norm": 3.4838390350341797,
|
| 1218 |
+
"learning_rate": 0.00019885042105564717,
|
| 1219 |
+
"loss": 5.6753,
|
| 1220 |
+
"step": 8650
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.1,
|
| 1224 |
+
"grad_norm": 4.251760959625244,
|
| 1225 |
+
"learning_rate": 0.00019882253394247381,
|
| 1226 |
+
"loss": 5.6303,
|
| 1227 |
+
"step": 8700
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.1,
|
| 1231 |
+
"grad_norm": 4.042376518249512,
|
| 1232 |
+
"learning_rate": 0.00019879431462315025,
|
| 1233 |
+
"loss": 5.5753,
|
| 1234 |
+
"step": 8750
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.1,
|
| 1238 |
+
"grad_norm": 4.239652633666992,
|
| 1239 |
+
"learning_rate": 0.0001987657631925398,
|
| 1240 |
+
"loss": 5.5335,
|
| 1241 |
+
"step": 8800
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.1,
|
| 1245 |
+
"grad_norm": 5.15481424331665,
|
| 1246 |
+
"learning_rate": 0.00019873687974662215,
|
| 1247 |
+
"loss": 5.5396,
|
| 1248 |
+
"step": 8850
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 0.1,
|
| 1252 |
+
"grad_norm": 4.36835241317749,
|
| 1253 |
+
"learning_rate": 0.00019870766438249317,
|
| 1254 |
+
"loss": 5.6017,
|
| 1255 |
+
"step": 8900
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.1,
|
| 1259 |
+
"grad_norm": 4.165258407592773,
|
| 1260 |
+
"learning_rate": 0.00019867811719836452,
|
| 1261 |
+
"loss": 5.7228,
|
| 1262 |
+
"step": 8950
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.1,
|
| 1266 |
+
"grad_norm": 4.125988006591797,
|
| 1267 |
+
"learning_rate": 0.0001986482382935633,
|
| 1268 |
+
"loss": 5.5787,
|
| 1269 |
+
"step": 9000
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.1,
|
| 1273 |
+
"grad_norm": 4.177731037139893,
|
| 1274 |
+
"learning_rate": 0.0001986180277685317,
|
| 1275 |
+
"loss": 5.5829,
|
| 1276 |
+
"step": 9050
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.1,
|
| 1280 |
+
"grad_norm": 5.006561279296875,
|
| 1281 |
+
"learning_rate": 0.00019858748572482683,
|
| 1282 |
+
"loss": 5.5466,
|
| 1283 |
+
"step": 9100
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.1,
|
| 1287 |
+
"grad_norm": 4.33070182800293,
|
| 1288 |
+
"learning_rate": 0.00019855661226512007,
|
| 1289 |
+
"loss": 5.5544,
|
| 1290 |
+
"step": 9150
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.1,
|
| 1294 |
+
"grad_norm": 4.358560085296631,
|
| 1295 |
+
"learning_rate": 0.00019852540749319708,
|
| 1296 |
+
"loss": 5.4599,
|
| 1297 |
+
"step": 9200
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 0.1,
|
| 1301 |
+
"grad_norm": 4.536096096038818,
|
| 1302 |
+
"learning_rate": 0.00019849387151395708,
|
| 1303 |
+
"loss": 5.4983,
|
| 1304 |
+
"step": 9250
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.1,
|
| 1308 |
+
"grad_norm": 4.66163444519043,
|
| 1309 |
+
"learning_rate": 0.0001984620044334129,
|
| 1310 |
+
"loss": 5.4097,
|
| 1311 |
+
"step": 9300
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.1,
|
| 1315 |
+
"grad_norm": 4.4319233894348145,
|
| 1316 |
+
"learning_rate": 0.00019842980635869024,
|
| 1317 |
+
"loss": 5.4093,
|
| 1318 |
+
"step": 9350
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 0.1,
|
| 1322 |
+
"grad_norm": 4.98419713973999,
|
| 1323 |
+
"learning_rate": 0.0001983972773980276,
|
| 1324 |
+
"loss": 5.4056,
|
| 1325 |
+
"step": 9400
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.1,
|
| 1329 |
+
"grad_norm": 3.6354339122772217,
|
| 1330 |
+
"learning_rate": 0.0001983644176607757,
|
| 1331 |
+
"loss": 5.3171,
|
| 1332 |
+
"step": 9450
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 0.11,
|
| 1336 |
+
"grad_norm": 4.495342254638672,
|
| 1337 |
+
"learning_rate": 0.00019833122725739736,
|
| 1338 |
+
"loss": 5.4521,
|
| 1339 |
+
"step": 9500
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 0.11,
|
| 1343 |
+
"grad_norm": 4.5558671951293945,
|
| 1344 |
+
"learning_rate": 0.00019829770629946678,
|
| 1345 |
+
"loss": 5.5158,
|
| 1346 |
+
"step": 9550
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.11,
|
| 1350 |
+
"grad_norm": 3.7165732383728027,
|
| 1351 |
+
"learning_rate": 0.00019826385489966957,
|
| 1352 |
+
"loss": 5.301,
|
| 1353 |
+
"step": 9600
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.11,
|
| 1357 |
+
"grad_norm": 6.030915260314941,
|
| 1358 |
+
"learning_rate": 0.00019822967317180204,
|
| 1359 |
+
"loss": 5.3316,
|
| 1360 |
+
"step": 9650
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.11,
|
| 1364 |
+
"grad_norm": 5.385923385620117,
|
| 1365 |
+
"learning_rate": 0.00019819516123077094,
|
| 1366 |
+
"loss": 5.3844,
|
| 1367 |
+
"step": 9700
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.11,
|
| 1371 |
+
"grad_norm": 4.383516788482666,
|
| 1372 |
+
"learning_rate": 0.00019816101926755305,
|
| 1373 |
+
"loss": 5.2995,
|
| 1374 |
+
"step": 9750
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.11,
|
| 1378 |
+
"grad_norm": 4.446406364440918,
|
| 1379 |
+
"learning_rate": 0.00019812585384780055,
|
| 1380 |
+
"loss": 5.386,
|
| 1381 |
+
"step": 9800
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 0.11,
|
| 1385 |
+
"grad_norm": 4.345483303070068,
|
| 1386 |
+
"learning_rate": 0.00019809035856388805,
|
| 1387 |
+
"loss": 5.2815,
|
| 1388 |
+
"step": 9850
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.11,
|
| 1392 |
+
"grad_norm": 4.791261672973633,
|
| 1393 |
+
"learning_rate": 0.00019805453353513813,
|
| 1394 |
+
"loss": 5.3757,
|
| 1395 |
+
"step": 9900
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.11,
|
| 1399 |
+
"grad_norm": 5.622151851654053,
|
| 1400 |
+
"learning_rate": 0.00019801837888198172,
|
| 1401 |
+
"loss": 5.4405,
|
| 1402 |
+
"step": 9950
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.11,
|
| 1406 |
+
"grad_norm": 4.934606075286865,
|
| 1407 |
+
"learning_rate": 0.0001979818947259579,
|
| 1408 |
+
"loss": 5.139,
|
| 1409 |
+
"step": 10000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 0.11,
|
| 1413 |
+
"grad_norm": 3.9659693241119385,
|
| 1414 |
+
"learning_rate": 0.0001979450811897134,
|
| 1415 |
+
"loss": 5.1726,
|
| 1416 |
+
"step": 10050
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 0.11,
|
| 1420 |
+
"grad_norm": 5.214992046356201,
|
| 1421 |
+
"learning_rate": 0.00019790793839700226,
|
| 1422 |
+
"loss": 5.2864,
|
| 1423 |
+
"step": 10100
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 0.11,
|
| 1427 |
+
"grad_norm": 4.5359601974487305,
|
| 1428 |
+
"learning_rate": 0.00019787046647268524,
|
| 1429 |
+
"loss": 5.1443,
|
| 1430 |
+
"step": 10150
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.11,
|
| 1434 |
+
"grad_norm": 4.26462984085083,
|
| 1435 |
+
"learning_rate": 0.00019783266554272962,
|
| 1436 |
+
"loss": 5.0597,
|
| 1437 |
+
"step": 10200
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.11,
|
| 1441 |
+
"grad_norm": 5.053945064544678,
|
| 1442 |
+
"learning_rate": 0.00019779453573420873,
|
| 1443 |
+
"loss": 5.2946,
|
| 1444 |
+
"step": 10250
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 0.11,
|
| 1448 |
+
"grad_norm": 6.082211494445801,
|
| 1449 |
+
"learning_rate": 0.00019775607717530127,
|
| 1450 |
+
"loss": 5.2075,
|
| 1451 |
+
"step": 10300
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 0.11,
|
| 1455 |
+
"grad_norm": 4.107390403747559,
|
| 1456 |
+
"learning_rate": 0.00019771728999529132,
|
| 1457 |
+
"loss": 5.1394,
|
| 1458 |
+
"step": 10350
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 0.12,
|
| 1462 |
+
"grad_norm": 4.58411169052124,
|
| 1463 |
+
"learning_rate": 0.00019767817432456752,
|
| 1464 |
+
"loss": 5.1064,
|
| 1465 |
+
"step": 10400
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.12,
|
| 1469 |
+
"grad_norm": 8.38965892791748,
|
| 1470 |
+
"learning_rate": 0.00019763952239228627,
|
| 1471 |
+
"loss": 5.0808,
|
| 1472 |
+
"step": 10450
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.12,
|
| 1476 |
+
"grad_norm": 3.885803699493408,
|
| 1477 |
+
"learning_rate": 0.00019759975669894338,
|
| 1478 |
+
"loss": 5.0664,
|
| 1479 |
+
"step": 10500
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.12,
|
| 1483 |
+
"grad_norm": 4.1605916023254395,
|
| 1484 |
+
"learning_rate": 0.00019755966290999167,
|
| 1485 |
+
"loss": 5.2469,
|
| 1486 |
+
"step": 10550
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 0.12,
|
| 1490 |
+
"grad_norm": 4.821887016296387,
|
| 1491 |
+
"learning_rate": 0.00019751924116021225,
|
| 1492 |
+
"loss": 5.2451,
|
| 1493 |
+
"step": 10600
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 0.12,
|
| 1497 |
+
"grad_norm": 3.865694761276245,
|
| 1498 |
+
"learning_rate": 0.00019747849158548858,
|
| 1499 |
+
"loss": 5.2334,
|
| 1500 |
+
"step": 10650
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 0.12,
|
| 1504 |
+
"grad_norm": 3.640681028366089,
|
| 1505 |
+
"learning_rate": 0.00019743741432280625,
|
| 1506 |
+
"loss": 5.1206,
|
| 1507 |
+
"step": 10700
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.12,
|
| 1511 |
+
"grad_norm": 4.04166316986084,
|
| 1512 |
+
"learning_rate": 0.00019739600951025236,
|
| 1513 |
+
"loss": 5.0059,
|
| 1514 |
+
"step": 10750
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.12,
|
| 1518 |
+
"grad_norm": 4.637605667114258,
|
| 1519 |
+
"learning_rate": 0.00019735427728701516,
|
| 1520 |
+
"loss": 5.0302,
|
| 1521 |
+
"step": 10800
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.12,
|
| 1525 |
+
"grad_norm": 4.08723783493042,
|
| 1526 |
+
"learning_rate": 0.0001973122177933835,
|
| 1527 |
+
"loss": 5.1551,
|
| 1528 |
+
"step": 10850
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 0.12,
|
| 1532 |
+
"grad_norm": 3.7944953441619873,
|
| 1533 |
+
"learning_rate": 0.00019726983117074643,
|
| 1534 |
+
"loss": 5.0665,
|
| 1535 |
+
"step": 10900
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 0.12,
|
| 1539 |
+
"grad_norm": 5.2847371101379395,
|
| 1540 |
+
"learning_rate": 0.00019722711756159266,
|
| 1541 |
+
"loss": 5.2212,
|
| 1542 |
+
"step": 10950
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 0.12,
|
| 1546 |
+
"grad_norm": 4.109150409698486,
|
| 1547 |
+
"learning_rate": 0.00019718407710951012,
|
| 1548 |
+
"loss": 5.2645,
|
| 1549 |
+
"step": 11000
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 0.12,
|
| 1553 |
+
"grad_norm": 4.127768039703369,
|
| 1554 |
+
"learning_rate": 0.0001971407099591855,
|
| 1555 |
+
"loss": 5.0395,
|
| 1556 |
+
"step": 11050
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.12,
|
| 1560 |
+
"grad_norm": 5.058667182922363,
|
| 1561 |
+
"learning_rate": 0.00019709701625640367,
|
| 1562 |
+
"loss": 5.0247,
|
| 1563 |
+
"step": 11100
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.12,
|
| 1567 |
+
"grad_norm": 5.4407267570495605,
|
| 1568 |
+
"learning_rate": 0.00019705299614804732,
|
| 1569 |
+
"loss": 4.9935,
|
| 1570 |
+
"step": 11150
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 0.12,
|
| 1574 |
+
"grad_norm": 3.7877707481384277,
|
| 1575 |
+
"learning_rate": 0.00019700864978209636,
|
| 1576 |
+
"loss": 5.074,
|
| 1577 |
+
"step": 11200
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 0.12,
|
| 1581 |
+
"grad_norm": 3.777330160140991,
|
| 1582 |
+
"learning_rate": 0.00019696397730762746,
|
| 1583 |
+
"loss": 5.0458,
|
| 1584 |
+
"step": 11250
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 0.13,
|
| 1588 |
+
"grad_norm": 4.143067836761475,
|
| 1589 |
+
"learning_rate": 0.0001969189788748136,
|
| 1590 |
+
"loss": 4.9375,
|
| 1591 |
+
"step": 11300
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 0.13,
|
| 1595 |
+
"grad_norm": 5.560107231140137,
|
| 1596 |
+
"learning_rate": 0.00019687365463492344,
|
| 1597 |
+
"loss": 4.8285,
|
| 1598 |
+
"step": 11350
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.13,
|
| 1602 |
+
"grad_norm": 4.057905197143555,
|
| 1603 |
+
"learning_rate": 0.00019682800474032095,
|
| 1604 |
+
"loss": 4.9753,
|
| 1605 |
+
"step": 11400
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.13,
|
| 1609 |
+
"grad_norm": 3.835442066192627,
|
| 1610 |
+
"learning_rate": 0.00019678202934446482,
|
| 1611 |
+
"loss": 4.9368,
|
| 1612 |
+
"step": 11450
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 0.13,
|
| 1616 |
+
"grad_norm": 5.135551929473877,
|
| 1617 |
+
"learning_rate": 0.0001967357286019079,
|
| 1618 |
+
"loss": 4.9994,
|
| 1619 |
+
"step": 11500
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 0.13,
|
| 1623 |
+
"grad_norm": 4.615053653717041,
|
| 1624 |
+
"learning_rate": 0.00019668910266829685,
|
| 1625 |
+
"loss": 5.0182,
|
| 1626 |
+
"step": 11550
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 0.13,
|
| 1630 |
+
"grad_norm": 4.474258899688721,
|
| 1631 |
+
"learning_rate": 0.0001966421517003714,
|
| 1632 |
+
"loss": 4.8704,
|
| 1633 |
+
"step": 11600
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 0.13,
|
| 1637 |
+
"grad_norm": 4.264945030212402,
|
| 1638 |
+
"learning_rate": 0.00019659487585596406,
|
| 1639 |
+
"loss": 4.9076,
|
| 1640 |
+
"step": 11650
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.13,
|
| 1644 |
+
"grad_norm": 4.091209411621094,
|
| 1645 |
+
"learning_rate": 0.00019654727529399925,
|
| 1646 |
+
"loss": 4.7135,
|
| 1647 |
+
"step": 11700
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.13,
|
| 1651 |
+
"grad_norm": 4.154038429260254,
|
| 1652 |
+
"learning_rate": 0.00019649935017449318,
|
| 1653 |
+
"loss": 4.8239,
|
| 1654 |
+
"step": 11750
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 0.13,
|
| 1658 |
+
"grad_norm": 3.697162628173828,
|
| 1659 |
+
"learning_rate": 0.00019645110065855305,
|
| 1660 |
+
"loss": 4.9972,
|
| 1661 |
+
"step": 11800
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 0.13,
|
| 1665 |
+
"grad_norm": 4.0024847984313965,
|
| 1666 |
+
"learning_rate": 0.00019640252690837645,
|
| 1667 |
+
"loss": 4.8854,
|
| 1668 |
+
"step": 11850
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 0.13,
|
| 1672 |
+
"grad_norm": 3.9416885375976562,
|
| 1673 |
+
"learning_rate": 0.0001963536290872511,
|
| 1674 |
+
"loss": 4.8547,
|
| 1675 |
+
"step": 11900
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 0.13,
|
| 1679 |
+
"grad_norm": 3.978651285171509,
|
| 1680 |
+
"learning_rate": 0.000196304407359554,
|
| 1681 |
+
"loss": 4.7873,
|
| 1682 |
+
"step": 11950
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 0.13,
|
| 1686 |
+
"grad_norm": 4.435175895690918,
|
| 1687 |
+
"learning_rate": 0.0001962548618907511,
|
| 1688 |
+
"loss": 4.8124,
|
| 1689 |
+
"step": 12000
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 0.13,
|
| 1693 |
+
"grad_norm": 3.8776824474334717,
|
| 1694 |
+
"learning_rate": 0.00019620499284739662,
|
| 1695 |
+
"loss": 4.8896,
|
| 1696 |
+
"step": 12050
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 0.13,
|
| 1700 |
+
"grad_norm": 5.041496276855469,
|
| 1701 |
+
"learning_rate": 0.00019615480039713248,
|
| 1702 |
+
"loss": 4.8343,
|
| 1703 |
+
"step": 12100
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 0.13,
|
| 1707 |
+
"grad_norm": 4.18281888961792,
|
| 1708 |
+
"learning_rate": 0.00019610428470868784,
|
| 1709 |
+
"loss": 4.8559,
|
| 1710 |
+
"step": 12150
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 0.14,
|
| 1714 |
+
"grad_norm": 4.223630905151367,
|
| 1715 |
+
"learning_rate": 0.00019605344595187844,
|
| 1716 |
+
"loss": 4.8153,
|
| 1717 |
+
"step": 12200
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 0.14,
|
| 1721 |
+
"grad_norm": 4.63677453994751,
|
| 1722 |
+
"learning_rate": 0.0001960022842976061,
|
| 1723 |
+
"loss": 4.7951,
|
| 1724 |
+
"step": 12250
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 0.14,
|
| 1728 |
+
"grad_norm": 4.188296794891357,
|
| 1729 |
+
"learning_rate": 0.00019595079991785802,
|
| 1730 |
+
"loss": 4.8904,
|
| 1731 |
+
"step": 12300
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 0.14,
|
| 1735 |
+
"grad_norm": 4.402559280395508,
|
| 1736 |
+
"learning_rate": 0.00019589899298570634,
|
| 1737 |
+
"loss": 4.7851,
|
| 1738 |
+
"step": 12350
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.14,
|
| 1742 |
+
"grad_norm": 5.976877212524414,
|
| 1743 |
+
"learning_rate": 0.00019584686367530755,
|
| 1744 |
+
"loss": 4.6431,
|
| 1745 |
+
"step": 12400
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 0.14,
|
| 1749 |
+
"grad_norm": 4.849298477172852,
|
| 1750 |
+
"learning_rate": 0.0001957944121619018,
|
| 1751 |
+
"loss": 4.7544,
|
| 1752 |
+
"step": 12450
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 0.14,
|
| 1756 |
+
"grad_norm": 4.932714462280273,
|
| 1757 |
+
"learning_rate": 0.0001957416386218124,
|
| 1758 |
+
"loss": 4.6811,
|
| 1759 |
+
"step": 12500
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 0.14,
|
| 1763 |
+
"grad_norm": 4.682474136352539,
|
| 1764 |
+
"learning_rate": 0.00019568854323244515,
|
| 1765 |
+
"loss": 4.799,
|
| 1766 |
+
"step": 12550
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 0.14,
|
| 1770 |
+
"grad_norm": 5.228520393371582,
|
| 1771 |
+
"learning_rate": 0.00019563619766470511,
|
| 1772 |
+
"loss": 4.7622,
|
| 1773 |
+
"step": 12600
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 0.14,
|
| 1777 |
+
"grad_norm": 4.093870162963867,
|
| 1778 |
+
"learning_rate": 0.00019558246554138458,
|
| 1779 |
+
"loss": 4.7369,
|
| 1780 |
+
"step": 12650
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 0.14,
|
| 1784 |
+
"grad_norm": 5.248356342315674,
|
| 1785 |
+
"learning_rate": 0.0001955284121038694,
|
| 1786 |
+
"loss": 4.7519,
|
| 1787 |
+
"step": 12700
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 0.14,
|
| 1791 |
+
"grad_norm": 3.924299955368042,
|
| 1792 |
+
"learning_rate": 0.00019547403753386803,
|
| 1793 |
+
"loss": 4.6441,
|
| 1794 |
+
"step": 12750
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 0.14,
|
| 1798 |
+
"grad_norm": 4.972569942474365,
|
| 1799 |
+
"learning_rate": 0.00019542043906868188,
|
| 1800 |
+
"loss": 4.7192,
|
| 1801 |
+
"step": 12800
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 0.14,
|
| 1805 |
+
"grad_norm": 5.033604145050049,
|
| 1806 |
+
"learning_rate": 0.00019536542919665846,
|
| 1807 |
+
"loss": 4.6397,
|
| 1808 |
+
"step": 12850
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 0.14,
|
| 1812 |
+
"grad_norm": 5.222695350646973,
|
| 1813 |
+
"learning_rate": 0.00019531009874003928,
|
| 1814 |
+
"loss": 4.6309,
|
| 1815 |
+
"step": 12900
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 0.14,
|
| 1819 |
+
"grad_norm": 3.810999631881714,
|
| 1820 |
+
"learning_rate": 0.00019525444788482562,
|
| 1821 |
+
"loss": 4.6513,
|
| 1822 |
+
"step": 12950
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 0.14,
|
| 1826 |
+
"grad_norm": 5.272600173950195,
|
| 1827 |
+
"learning_rate": 0.00019519847681809585,
|
| 1828 |
+
"loss": 4.8001,
|
| 1829 |
+
"step": 13000
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 0.14,
|
| 1833 |
+
"grad_norm": 4.836308002471924,
|
| 1834 |
+
"learning_rate": 0.00019514218572800468,
|
| 1835 |
+
"loss": 4.7101,
|
| 1836 |
+
"step": 13050
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 0.15,
|
| 1840 |
+
"grad_norm": 4.598148345947266,
|
| 1841 |
+
"learning_rate": 0.00019508557480378276,
|
| 1842 |
+
"loss": 4.5578,
|
| 1843 |
+
"step": 13100
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 0.15,
|
| 1847 |
+
"grad_norm": 3.910820722579956,
|
| 1848 |
+
"learning_rate": 0.0001950286442357358,
|
| 1849 |
+
"loss": 4.7124,
|
| 1850 |
+
"step": 13150
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.15,
|
| 1854 |
+
"grad_norm": 3.856081962585449,
|
| 1855 |
+
"learning_rate": 0.00019497139421524416,
|
| 1856 |
+
"loss": 4.7563,
|
| 1857 |
+
"step": 13200
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 0.15,
|
| 1861 |
+
"grad_norm": 4.151907920837402,
|
| 1862 |
+
"learning_rate": 0.00019491382493476195,
|
| 1863 |
+
"loss": 4.6726,
|
| 1864 |
+
"step": 13250
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 0.15,
|
| 1868 |
+
"grad_norm": 4.349935054779053,
|
| 1869 |
+
"learning_rate": 0.0001948559365878166,
|
| 1870 |
+
"loss": 4.6341,
|
| 1871 |
+
"step": 13300
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 0.15,
|
| 1875 |
+
"grad_norm": 3.8229756355285645,
|
| 1876 |
+
"learning_rate": 0.00019479772936900811,
|
| 1877 |
+
"loss": 4.6183,
|
| 1878 |
+
"step": 13350
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 0.15,
|
| 1882 |
+
"grad_norm": 5.495506286621094,
|
| 1883 |
+
"learning_rate": 0.0001947392034740084,
|
| 1884 |
+
"loss": 4.6608,
|
| 1885 |
+
"step": 13400
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 0.15,
|
| 1889 |
+
"grad_norm": 4.307513236999512,
|
| 1890 |
+
"learning_rate": 0.00019468035909956072,
|
| 1891 |
+
"loss": 4.6805,
|
| 1892 |
+
"step": 13450
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 0.15,
|
| 1896 |
+
"grad_norm": 3.939659595489502,
|
| 1897 |
+
"learning_rate": 0.0001946211964434788,
|
| 1898 |
+
"loss": 4.679,
|
| 1899 |
+
"step": 13500
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 0.15,
|
| 1903 |
+
"grad_norm": 5.444967269897461,
|
| 1904 |
+
"learning_rate": 0.00019456171570464653,
|
| 1905 |
+
"loss": 4.7195,
|
| 1906 |
+
"step": 13550
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 0.15,
|
| 1910 |
+
"grad_norm": 4.513270854949951,
|
| 1911 |
+
"learning_rate": 0.00019450191708301687,
|
| 1912 |
+
"loss": 4.5367,
|
| 1913 |
+
"step": 13600
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 0.15,
|
| 1917 |
+
"grad_norm": 4.617405414581299,
|
| 1918 |
+
"learning_rate": 0.00019444180077961146,
|
| 1919 |
+
"loss": 4.5742,
|
| 1920 |
+
"step": 13650
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 0.15,
|
| 1924 |
+
"grad_norm": 4.580646991729736,
|
| 1925 |
+
"learning_rate": 0.00019438136699652001,
|
| 1926 |
+
"loss": 4.4936,
|
| 1927 |
+
"step": 13700
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 0.15,
|
| 1931 |
+
"grad_norm": 4.657532691955566,
|
| 1932 |
+
"learning_rate": 0.00019432061593689927,
|
| 1933 |
+
"loss": 4.6877,
|
| 1934 |
+
"step": 13750
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 0.15,
|
| 1938 |
+
"grad_norm": 5.374803066253662,
|
| 1939 |
+
"learning_rate": 0.0001942595478049727,
|
| 1940 |
+
"loss": 4.6101,
|
| 1941 |
+
"step": 13800
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 0.15,
|
| 1945 |
+
"grad_norm": 5.1111650466918945,
|
| 1946 |
+
"learning_rate": 0.00019419816280602962,
|
| 1947 |
+
"loss": 4.6185,
|
| 1948 |
+
"step": 13850
|
| 1949 |
+
},
|
| 1950 |
+
{
|
| 1951 |
+
"epoch": 0.15,
|
| 1952 |
+
"grad_norm": 5.18306303024292,
|
| 1953 |
+
"learning_rate": 0.00019413646114642446,
|
| 1954 |
+
"loss": 4.5524,
|
| 1955 |
+
"step": 13900
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 0.15,
|
| 1959 |
+
"grad_norm": 4.411191463470459,
|
| 1960 |
+
"learning_rate": 0.00019407444303357624,
|
| 1961 |
+
"loss": 4.4346,
|
| 1962 |
+
"step": 13950
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 0.16,
|
| 1966 |
+
"grad_norm": 4.161925792694092,
|
| 1967 |
+
"learning_rate": 0.0001940121086759678,
|
| 1968 |
+
"loss": 4.3702,
|
| 1969 |
+
"step": 14000
|
| 1970 |
+
}
|
| 1971 |
+
],
|
| 1972 |
+
"logging_steps": 50,
|
| 1973 |
+
"max_steps": 90183,
|
| 1974 |
+
"num_input_tokens_seen": 0,
|
| 1975 |
+
"num_train_epochs": 1,
|
| 1976 |
+
"save_steps": 1000,
|
| 1977 |
+
"total_flos": 5034267377664000.0,
|
| 1978 |
+
"train_batch_size": 16,
|
| 1979 |
+
"trial_name": null,
|
| 1980 |
+
"trial_params": null
|
| 1981 |
+
}
|
checkpoints/checkpoint-14000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bcdffec292848a9e4baadecd0be02bf6760a1c3238602b1aad355cf1ba43d36e
|
| 3 |
+
size 5048
|
checkpoints/config.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"AlbertForMaskedLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0,
|
| 6 |
+
"bos_token_id": 2,
|
| 7 |
+
"classifier_dropout_prob": 0.1,
|
| 8 |
+
"embedding_size": 128,
|
| 9 |
+
"eos_token_id": 3,
|
| 10 |
+
"hidden_act": "gelu_new",
|
| 11 |
+
"hidden_dropout_prob": 0,
|
| 12 |
+
"hidden_size": 768,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"inner_group_num": 1,
|
| 15 |
+
"intermediate_size": 3072,
|
| 16 |
+
"layer_norm_eps": 1e-12,
|
| 17 |
+
"max_position_embeddings": 512,
|
| 18 |
+
"model_type": "albert",
|
| 19 |
+
"num_attention_heads": 12,
|
| 20 |
+
"num_hidden_groups": 1,
|
| 21 |
+
"num_hidden_layers": 12,
|
| 22 |
+
"pad_token_id": 0,
|
| 23 |
+
"position_embedding_type": "absolute",
|
| 24 |
+
"torch_dtype": "float32",
|
| 25 |
+
"transformers_version": "4.38.2",
|
| 26 |
+
"type_vocab_size": 2,
|
| 27 |
+
"vocab_size": 30000
|
| 28 |
+
}
|
checkpoints/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5e0a3efb18a87263970f630bf8170263f502b54358886171fe7e3c20a2f85321
|
| 3 |
+
size 44890256
|
checkpoints/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bcdffec292848a9e4baadecd0be02bf6760a1c3238602b1aad355cf1ba43d36e
|
| 3 |
+
size 5048
|