Upload folder using huggingface_hub
Browse files- config.json +26 -0
- optimizer.pt +3 -0
- pytorch_model.bin +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +1661 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"RobertaForMaskedLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.1,
|
| 6 |
+
"bos_token_id": 0,
|
| 7 |
+
"classifier_dropout": null,
|
| 8 |
+
"eos_token_id": 2,
|
| 9 |
+
"hidden_act": "gelu",
|
| 10 |
+
"hidden_dropout_prob": 0.1,
|
| 11 |
+
"hidden_size": 768,
|
| 12 |
+
"initializer_range": 0.02,
|
| 13 |
+
"intermediate_size": 3072,
|
| 14 |
+
"layer_norm_eps": 1e-12,
|
| 15 |
+
"max_position_embeddings": 514,
|
| 16 |
+
"model_type": "roberta",
|
| 17 |
+
"num_attention_heads": 12,
|
| 18 |
+
"num_hidden_layers": 12,
|
| 19 |
+
"pad_token_id": 1,
|
| 20 |
+
"position_embedding_type": "absolute",
|
| 21 |
+
"torch_dtype": "float32",
|
| 22 |
+
"transformers_version": "4.45.2",
|
| 23 |
+
"type_vocab_size": 2,
|
| 24 |
+
"use_cache": true,
|
| 25 |
+
"vocab_size": 50265
|
| 26 |
+
}
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ecdacc48f5300b997d91affc69ff1027a0bd529b84cf1575e414f849064818d8
|
| 3 |
+
size 997754490
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4d4030ded4be4516c6f6aa6112a80c1a44dc0250e9ea7dfe91ae55b23c058cf7
|
| 3 |
+
size 498862706
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:71b7219197bbf6fad348242545bcc0e627442a97a07d6f0a625924b0335f95d4
|
| 3 |
+
size 14244
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bdd055ee3fb8747b8f17aafdd48cab2c33b4e1c6b17022fd73bb72c87cb96559
|
| 3 |
+
size 1064
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1661 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 1.4024385213851929,
|
| 3 |
+
"best_model_checkpoint": "./outputs/202410/no_safetensors/checkpoint/checkpoint-110000",
|
| 4 |
+
"epoch": 3.5091956039749252,
|
| 5 |
+
"eval_steps": 10000,
|
| 6 |
+
"global_step": 110000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.01595074410221237,
|
| 13 |
+
"grad_norm": 108751.9921875,
|
| 14 |
+
"learning_rate": 5e-06,
|
| 15 |
+
"loss": 9.707,
|
| 16 |
+
"step": 500
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.03190148820442474,
|
| 20 |
+
"grad_norm": 44635.21484375,
|
| 21 |
+
"learning_rate": 1e-05,
|
| 22 |
+
"loss": 7.8592,
|
| 23 |
+
"step": 1000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.047852232306637106,
|
| 27 |
+
"grad_norm": 56782.453125,
|
| 28 |
+
"learning_rate": 1.5e-05,
|
| 29 |
+
"loss": 7.2609,
|
| 30 |
+
"step": 1500
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.06380297640884948,
|
| 34 |
+
"grad_norm": 96195.9140625,
|
| 35 |
+
"learning_rate": 2e-05,
|
| 36 |
+
"loss": 7.1215,
|
| 37 |
+
"step": 2000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.07975372051106185,
|
| 41 |
+
"grad_norm": 108495.3828125,
|
| 42 |
+
"learning_rate": 2.5e-05,
|
| 43 |
+
"loss": 7.0039,
|
| 44 |
+
"step": 2500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.09570446461327421,
|
| 48 |
+
"grad_norm": 69317.0625,
|
| 49 |
+
"learning_rate": 3e-05,
|
| 50 |
+
"loss": 6.8924,
|
| 51 |
+
"step": 3000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.11165520871548658,
|
| 55 |
+
"grad_norm": 112126.8203125,
|
| 56 |
+
"learning_rate": 3.5e-05,
|
| 57 |
+
"loss": 6.7924,
|
| 58 |
+
"step": 3500
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.12760595281769896,
|
| 62 |
+
"grad_norm": 94684.1640625,
|
| 63 |
+
"learning_rate": 4e-05,
|
| 64 |
+
"loss": 6.6991,
|
| 65 |
+
"step": 4000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.14355669691991133,
|
| 69 |
+
"grad_norm": 99549.90625,
|
| 70 |
+
"learning_rate": 4.5e-05,
|
| 71 |
+
"loss": 6.6077,
|
| 72 |
+
"step": 4500
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.1595074410221237,
|
| 76 |
+
"grad_norm": 100152.6171875,
|
| 77 |
+
"learning_rate": 5e-05,
|
| 78 |
+
"loss": 6.4957,
|
| 79 |
+
"step": 5000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.17545818512433606,
|
| 83 |
+
"grad_norm": 165780.328125,
|
| 84 |
+
"learning_rate": 5.500000000000001e-05,
|
| 85 |
+
"loss": 6.3014,
|
| 86 |
+
"step": 5500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.19140892922654842,
|
| 90 |
+
"grad_norm": 132947.90625,
|
| 91 |
+
"learning_rate": 6e-05,
|
| 92 |
+
"loss": 5.9146,
|
| 93 |
+
"step": 6000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.2073596733287608,
|
| 97 |
+
"grad_norm": 146624.34375,
|
| 98 |
+
"learning_rate": 6.500000000000001e-05,
|
| 99 |
+
"loss": 5.5084,
|
| 100 |
+
"step": 6500
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.22331041743097316,
|
| 104 |
+
"grad_norm": 145966.1875,
|
| 105 |
+
"learning_rate": 7e-05,
|
| 106 |
+
"loss": 5.1444,
|
| 107 |
+
"step": 7000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.23926116153318552,
|
| 111 |
+
"grad_norm": 127509.96875,
|
| 112 |
+
"learning_rate": 7.500000000000001e-05,
|
| 113 |
+
"loss": 4.5906,
|
| 114 |
+
"step": 7500
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2552119056353979,
|
| 118 |
+
"grad_norm": 97949.8828125,
|
| 119 |
+
"learning_rate": 8e-05,
|
| 120 |
+
"loss": 3.9702,
|
| 121 |
+
"step": 8000
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.2711626497376103,
|
| 125 |
+
"grad_norm": 89478.4296875,
|
| 126 |
+
"learning_rate": 8.5e-05,
|
| 127 |
+
"loss": 3.5577,
|
| 128 |
+
"step": 8500
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.28711339383982265,
|
| 132 |
+
"grad_norm": 94187.8984375,
|
| 133 |
+
"learning_rate": 9e-05,
|
| 134 |
+
"loss": 3.302,
|
| 135 |
+
"step": 9000
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.303064137942035,
|
| 139 |
+
"grad_norm": 83401.4765625,
|
| 140 |
+
"learning_rate": 9.5e-05,
|
| 141 |
+
"loss": 3.122,
|
| 142 |
+
"step": 9500
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.3190148820442474,
|
| 146 |
+
"grad_norm": 80432.59375,
|
| 147 |
+
"learning_rate": 0.0001,
|
| 148 |
+
"loss": 2.9716,
|
| 149 |
+
"step": 10000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.3190148820442474,
|
| 153 |
+
"eval_loss": 2.7994558811187744,
|
| 154 |
+
"eval_runtime": 4941.5974,
|
| 155 |
+
"eval_samples_per_second": 202.986,
|
| 156 |
+
"eval_steps_per_second": 1.586,
|
| 157 |
+
"step": 10000
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.33496562614645975,
|
| 161 |
+
"grad_norm": 71825.7421875,
|
| 162 |
+
"learning_rate": 9.994949494949496e-05,
|
| 163 |
+
"loss": 2.8485,
|
| 164 |
+
"step": 10500
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.3509163702486721,
|
| 168 |
+
"grad_norm": 76352.0625,
|
| 169 |
+
"learning_rate": 9.98989898989899e-05,
|
| 170 |
+
"loss": 2.7498,
|
| 171 |
+
"step": 11000
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.3668671143508845,
|
| 175 |
+
"grad_norm": 74371.5703125,
|
| 176 |
+
"learning_rate": 9.984848484848486e-05,
|
| 177 |
+
"loss": 2.6733,
|
| 178 |
+
"step": 11500
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.38281785845309685,
|
| 182 |
+
"grad_norm": 68137.9609375,
|
| 183 |
+
"learning_rate": 9.97979797979798e-05,
|
| 184 |
+
"loss": 2.5987,
|
| 185 |
+
"step": 12000
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.3987686025553092,
|
| 189 |
+
"grad_norm": 64699.58984375,
|
| 190 |
+
"learning_rate": 9.974747474747475e-05,
|
| 191 |
+
"loss": 2.5405,
|
| 192 |
+
"step": 12500
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.4147193466575216,
|
| 196 |
+
"grad_norm": 69361.3046875,
|
| 197 |
+
"learning_rate": 9.96969696969697e-05,
|
| 198 |
+
"loss": 2.4881,
|
| 199 |
+
"step": 13000
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.43067009075973395,
|
| 203 |
+
"grad_norm": 69834.671875,
|
| 204 |
+
"learning_rate": 9.964646464646466e-05,
|
| 205 |
+
"loss": 2.446,
|
| 206 |
+
"step": 13500
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.4466208348619463,
|
| 210 |
+
"grad_norm": 57975.98828125,
|
| 211 |
+
"learning_rate": 9.95959595959596e-05,
|
| 212 |
+
"loss": 2.3991,
|
| 213 |
+
"step": 14000
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.4625715789641587,
|
| 217 |
+
"grad_norm": 67055.9375,
|
| 218 |
+
"learning_rate": 9.954545454545455e-05,
|
| 219 |
+
"loss": 2.3637,
|
| 220 |
+
"step": 14500
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.47852232306637105,
|
| 224 |
+
"grad_norm": 62929.25390625,
|
| 225 |
+
"learning_rate": 9.94949494949495e-05,
|
| 226 |
+
"loss": 2.3294,
|
| 227 |
+
"step": 15000
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.4944730671685834,
|
| 231 |
+
"grad_norm": 65925.8828125,
|
| 232 |
+
"learning_rate": 9.944444444444446e-05,
|
| 233 |
+
"loss": 2.3016,
|
| 234 |
+
"step": 15500
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.5104238112707958,
|
| 238 |
+
"grad_norm": 59559.88671875,
|
| 239 |
+
"learning_rate": 9.939393939393939e-05,
|
| 240 |
+
"loss": 2.2717,
|
| 241 |
+
"step": 16000
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.5263745553730081,
|
| 245 |
+
"grad_norm": 63280.40234375,
|
| 246 |
+
"learning_rate": 9.934343434343435e-05,
|
| 247 |
+
"loss": 2.2462,
|
| 248 |
+
"step": 16500
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.5423252994752206,
|
| 252 |
+
"grad_norm": 61513.0859375,
|
| 253 |
+
"learning_rate": 9.92929292929293e-05,
|
| 254 |
+
"loss": 2.2214,
|
| 255 |
+
"step": 17000
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.5582760435774329,
|
| 259 |
+
"grad_norm": 63240.92578125,
|
| 260 |
+
"learning_rate": 9.924242424242425e-05,
|
| 261 |
+
"loss": 2.2013,
|
| 262 |
+
"step": 17500
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.5742267876796453,
|
| 266 |
+
"grad_norm": 61634.92578125,
|
| 267 |
+
"learning_rate": 9.919191919191919e-05,
|
| 268 |
+
"loss": 2.1798,
|
| 269 |
+
"step": 18000
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.5901775317818576,
|
| 273 |
+
"grad_norm": 56590.8203125,
|
| 274 |
+
"learning_rate": 9.914141414141415e-05,
|
| 275 |
+
"loss": 2.1557,
|
| 276 |
+
"step": 18500
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.60612827588407,
|
| 280 |
+
"grad_norm": 59655.97265625,
|
| 281 |
+
"learning_rate": 9.909090909090911e-05,
|
| 282 |
+
"loss": 2.14,
|
| 283 |
+
"step": 19000
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6220790199862823,
|
| 287 |
+
"grad_norm": 60380.125,
|
| 288 |
+
"learning_rate": 9.904040404040404e-05,
|
| 289 |
+
"loss": 2.123,
|
| 290 |
+
"step": 19500
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.6380297640884948,
|
| 294 |
+
"grad_norm": 56893.875,
|
| 295 |
+
"learning_rate": 9.8989898989899e-05,
|
| 296 |
+
"loss": 2.1062,
|
| 297 |
+
"step": 20000
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.6380297640884948,
|
| 301 |
+
"eval_loss": 2.007786512374878,
|
| 302 |
+
"eval_runtime": 4985.2285,
|
| 303 |
+
"eval_samples_per_second": 201.209,
|
| 304 |
+
"eval_steps_per_second": 1.572,
|
| 305 |
+
"step": 20000
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.6539805081907071,
|
| 309 |
+
"grad_norm": 60312.33203125,
|
| 310 |
+
"learning_rate": 9.893939393939395e-05,
|
| 311 |
+
"loss": 2.0889,
|
| 312 |
+
"step": 20500
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.6699312522929195,
|
| 316 |
+
"grad_norm": 58157.28125,
|
| 317 |
+
"learning_rate": 9.888888888888889e-05,
|
| 318 |
+
"loss": 2.0776,
|
| 319 |
+
"step": 21000
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.6858819963951318,
|
| 323 |
+
"grad_norm": 58377.234375,
|
| 324 |
+
"learning_rate": 9.883838383838384e-05,
|
| 325 |
+
"loss": 2.0646,
|
| 326 |
+
"step": 21500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.7018327404973442,
|
| 330 |
+
"grad_norm": 57768.578125,
|
| 331 |
+
"learning_rate": 9.87878787878788e-05,
|
| 332 |
+
"loss": 2.0493,
|
| 333 |
+
"step": 22000
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.7177834845995565,
|
| 337 |
+
"grad_norm": 56948.41796875,
|
| 338 |
+
"learning_rate": 9.873737373737374e-05,
|
| 339 |
+
"loss": 2.0339,
|
| 340 |
+
"step": 22500
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.733734228701769,
|
| 344 |
+
"grad_norm": 58923.421875,
|
| 345 |
+
"learning_rate": 9.868686868686869e-05,
|
| 346 |
+
"loss": 2.0256,
|
| 347 |
+
"step": 23000
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.7496849728039813,
|
| 351 |
+
"grad_norm": 60381.23828125,
|
| 352 |
+
"learning_rate": 9.863636363636364e-05,
|
| 353 |
+
"loss": 2.0108,
|
| 354 |
+
"step": 23500
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.7656357169061937,
|
| 358 |
+
"grad_norm": 55685.12890625,
|
| 359 |
+
"learning_rate": 9.85858585858586e-05,
|
| 360 |
+
"loss": 2.0011,
|
| 361 |
+
"step": 24000
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.781586461008406,
|
| 365 |
+
"grad_norm": 55655.4609375,
|
| 366 |
+
"learning_rate": 9.853535353535353e-05,
|
| 367 |
+
"loss": 1.9899,
|
| 368 |
+
"step": 24500
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.7975372051106184,
|
| 372 |
+
"grad_norm": 57943.70703125,
|
| 373 |
+
"learning_rate": 9.848484848484849e-05,
|
| 374 |
+
"loss": 1.9793,
|
| 375 |
+
"step": 25000
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.8134879492128307,
|
| 379 |
+
"grad_norm": 56282.34375,
|
| 380 |
+
"learning_rate": 9.843434343434344e-05,
|
| 381 |
+
"loss": 1.9666,
|
| 382 |
+
"step": 25500
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.8294386933150432,
|
| 386 |
+
"grad_norm": 56147.5390625,
|
| 387 |
+
"learning_rate": 9.838383838383838e-05,
|
| 388 |
+
"loss": 1.9575,
|
| 389 |
+
"step": 26000
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.8453894374172555,
|
| 393 |
+
"grad_norm": 54766.921875,
|
| 394 |
+
"learning_rate": 9.833333333333333e-05,
|
| 395 |
+
"loss": 1.9508,
|
| 396 |
+
"step": 26500
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.8613401815194679,
|
| 400 |
+
"grad_norm": 57243.0390625,
|
| 401 |
+
"learning_rate": 9.828282828282829e-05,
|
| 402 |
+
"loss": 1.9406,
|
| 403 |
+
"step": 27000
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.8772909256216802,
|
| 407 |
+
"grad_norm": 57537.25,
|
| 408 |
+
"learning_rate": 9.823232323232325e-05,
|
| 409 |
+
"loss": 1.9295,
|
| 410 |
+
"step": 27500
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.8932416697238926,
|
| 414 |
+
"grad_norm": 56535.60546875,
|
| 415 |
+
"learning_rate": 9.818181818181818e-05,
|
| 416 |
+
"loss": 1.9232,
|
| 417 |
+
"step": 28000
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.9091924138261049,
|
| 421 |
+
"grad_norm": 60243.4375,
|
| 422 |
+
"learning_rate": 9.813131313131314e-05,
|
| 423 |
+
"loss": 1.9139,
|
| 424 |
+
"step": 28500
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.9251431579283174,
|
| 428 |
+
"grad_norm": 54113.95703125,
|
| 429 |
+
"learning_rate": 9.808080808080809e-05,
|
| 430 |
+
"loss": 1.9067,
|
| 431 |
+
"step": 29000
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.9410939020305297,
|
| 435 |
+
"grad_norm": 56388.44921875,
|
| 436 |
+
"learning_rate": 9.803030303030303e-05,
|
| 437 |
+
"loss": 1.8986,
|
| 438 |
+
"step": 29500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.9570446461327421,
|
| 442 |
+
"grad_norm": 55065.8203125,
|
| 443 |
+
"learning_rate": 9.797979797979798e-05,
|
| 444 |
+
"loss": 1.8912,
|
| 445 |
+
"step": 30000
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.9570446461327421,
|
| 449 |
+
"eval_loss": 1.7991323471069336,
|
| 450 |
+
"eval_runtime": 4913.8629,
|
| 451 |
+
"eval_samples_per_second": 204.132,
|
| 452 |
+
"eval_steps_per_second": 1.595,
|
| 453 |
+
"step": 30000
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.9729953902349545,
|
| 457 |
+
"grad_norm": 55303.02734375,
|
| 458 |
+
"learning_rate": 9.792929292929294e-05,
|
| 459 |
+
"loss": 1.8815,
|
| 460 |
+
"step": 30500
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 0.9889461343371668,
|
| 464 |
+
"grad_norm": 61240.37109375,
|
| 465 |
+
"learning_rate": 9.787878787878789e-05,
|
| 466 |
+
"loss": 1.876,
|
| 467 |
+
"step": 31000
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 1.0048968784393792,
|
| 471 |
+
"grad_norm": 54453.8359375,
|
| 472 |
+
"learning_rate": 9.782828282828283e-05,
|
| 473 |
+
"loss": 1.8677,
|
| 474 |
+
"step": 31500
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 1.0208476225415917,
|
| 478 |
+
"grad_norm": 53728.48828125,
|
| 479 |
+
"learning_rate": 9.777777777777778e-05,
|
| 480 |
+
"loss": 1.8582,
|
| 481 |
+
"step": 32000
|
| 482 |
+
},
|
| 483 |
+
{
|
| 484 |
+
"epoch": 1.0367983666438039,
|
| 485 |
+
"grad_norm": 57865.578125,
|
| 486 |
+
"learning_rate": 9.772727272727274e-05,
|
| 487 |
+
"loss": 1.8532,
|
| 488 |
+
"step": 32500
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 1.0527491107460163,
|
| 492 |
+
"grad_norm": 57046.4609375,
|
| 493 |
+
"learning_rate": 9.767676767676767e-05,
|
| 494 |
+
"loss": 1.8473,
|
| 495 |
+
"step": 33000
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 1.0686998548482287,
|
| 499 |
+
"grad_norm": 57173.7265625,
|
| 500 |
+
"learning_rate": 9.762626262626263e-05,
|
| 501 |
+
"loss": 1.8394,
|
| 502 |
+
"step": 33500
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 1.0846505989504411,
|
| 506 |
+
"grad_norm": 55721.00390625,
|
| 507 |
+
"learning_rate": 9.757575757575758e-05,
|
| 508 |
+
"loss": 1.8346,
|
| 509 |
+
"step": 34000
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 1.1006013430526533,
|
| 513 |
+
"grad_norm": 56324.05859375,
|
| 514 |
+
"learning_rate": 9.752525252525253e-05,
|
| 515 |
+
"loss": 1.8319,
|
| 516 |
+
"step": 34500
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 1.1165520871548658,
|
| 520 |
+
"grad_norm": 55807.09765625,
|
| 521 |
+
"learning_rate": 9.747474747474747e-05,
|
| 522 |
+
"loss": 1.8231,
|
| 523 |
+
"step": 35000
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 1.1325028312570782,
|
| 527 |
+
"grad_norm": 53290.69140625,
|
| 528 |
+
"learning_rate": 9.742424242424243e-05,
|
| 529 |
+
"loss": 1.815,
|
| 530 |
+
"step": 35500
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 1.1484535753592906,
|
| 534 |
+
"grad_norm": 55672.96875,
|
| 535 |
+
"learning_rate": 9.737373737373738e-05,
|
| 536 |
+
"loss": 1.8117,
|
| 537 |
+
"step": 36000
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 1.1644043194615028,
|
| 541 |
+
"grad_norm": 53640.85546875,
|
| 542 |
+
"learning_rate": 9.732323232323232e-05,
|
| 543 |
+
"loss": 1.8069,
|
| 544 |
+
"step": 36500
|
| 545 |
+
},
|
| 546 |
+
{
|
| 547 |
+
"epoch": 1.1803550635637152,
|
| 548 |
+
"grad_norm": 58046.828125,
|
| 549 |
+
"learning_rate": 9.727272727272728e-05,
|
| 550 |
+
"loss": 1.801,
|
| 551 |
+
"step": 37000
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 1.1963058076659276,
|
| 555 |
+
"grad_norm": 57318.890625,
|
| 556 |
+
"learning_rate": 9.722222222222223e-05,
|
| 557 |
+
"loss": 1.7943,
|
| 558 |
+
"step": 37500
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 1.21225655176814,
|
| 562 |
+
"grad_norm": 57652.68359375,
|
| 563 |
+
"learning_rate": 9.717171717171718e-05,
|
| 564 |
+
"loss": 1.7907,
|
| 565 |
+
"step": 38000
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 1.2282072958703523,
|
| 569 |
+
"grad_norm": 54140.81640625,
|
| 570 |
+
"learning_rate": 9.712121212121212e-05,
|
| 571 |
+
"loss": 1.7837,
|
| 572 |
+
"step": 38500
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 1.2441580399725647,
|
| 576 |
+
"grad_norm": 55963.265625,
|
| 577 |
+
"learning_rate": 9.707070707070708e-05,
|
| 578 |
+
"loss": 1.7804,
|
| 579 |
+
"step": 39000
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 1.260108784074777,
|
| 583 |
+
"grad_norm": 56717.23828125,
|
| 584 |
+
"learning_rate": 9.702020202020202e-05,
|
| 585 |
+
"loss": 1.7779,
|
| 586 |
+
"step": 39500
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.2760595281769895,
|
| 590 |
+
"grad_norm": 54491.91796875,
|
| 591 |
+
"learning_rate": 9.696969696969698e-05,
|
| 592 |
+
"loss": 1.7704,
|
| 593 |
+
"step": 40000
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.2760595281769895,
|
| 597 |
+
"eval_loss": 1.684906005859375,
|
| 598 |
+
"eval_runtime": 4951.5896,
|
| 599 |
+
"eval_samples_per_second": 202.576,
|
| 600 |
+
"eval_steps_per_second": 1.583,
|
| 601 |
+
"step": 40000
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 1.2920262230233042,
|
| 605 |
+
"grad_norm": 53837.6640625,
|
| 606 |
+
"learning_rate": 9.691919191919192e-05,
|
| 607 |
+
"loss": 1.7651,
|
| 608 |
+
"step": 40500
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 1.3079769671255164,
|
| 612 |
+
"grad_norm": 54601.4765625,
|
| 613 |
+
"learning_rate": 9.686868686868688e-05,
|
| 614 |
+
"loss": 1.759,
|
| 615 |
+
"step": 41000
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 1.3239277112277288,
|
| 619 |
+
"grad_norm": 52770.5078125,
|
| 620 |
+
"learning_rate": 9.681818181818181e-05,
|
| 621 |
+
"loss": 1.7568,
|
| 622 |
+
"step": 41500
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.3398784553299412,
|
| 626 |
+
"grad_norm": 56085.59765625,
|
| 627 |
+
"learning_rate": 9.676767676767677e-05,
|
| 628 |
+
"loss": 1.7522,
|
| 629 |
+
"step": 42000
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.3558291994321534,
|
| 633 |
+
"grad_norm": 55068.359375,
|
| 634 |
+
"learning_rate": 9.671717171717172e-05,
|
| 635 |
+
"loss": 1.7506,
|
| 636 |
+
"step": 42500
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.3717799435343658,
|
| 640 |
+
"grad_norm": 57091.0546875,
|
| 641 |
+
"learning_rate": 9.666666666666667e-05,
|
| 642 |
+
"loss": 1.7419,
|
| 643 |
+
"step": 43000
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.3877306876365783,
|
| 647 |
+
"grad_norm": 52158.63671875,
|
| 648 |
+
"learning_rate": 9.661616161616161e-05,
|
| 649 |
+
"loss": 1.7395,
|
| 650 |
+
"step": 43500
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.4036814317387907,
|
| 654 |
+
"grad_norm": 54727.2890625,
|
| 655 |
+
"learning_rate": 9.656565656565657e-05,
|
| 656 |
+
"loss": 1.7351,
|
| 657 |
+
"step": 44000
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.419632175841003,
|
| 661 |
+
"grad_norm": 59289.56640625,
|
| 662 |
+
"learning_rate": 9.651515151515152e-05,
|
| 663 |
+
"loss": 1.731,
|
| 664 |
+
"step": 44500
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 1.4355829199432153,
|
| 668 |
+
"grad_norm": 57592.5546875,
|
| 669 |
+
"learning_rate": 9.646464646464647e-05,
|
| 670 |
+
"loss": 1.7268,
|
| 671 |
+
"step": 45000
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 1.4515336640454277,
|
| 675 |
+
"grad_norm": 56016.44921875,
|
| 676 |
+
"learning_rate": 9.641414141414143e-05,
|
| 677 |
+
"loss": 1.723,
|
| 678 |
+
"step": 45500
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 1.4674844081476401,
|
| 682 |
+
"grad_norm": 55521.9765625,
|
| 683 |
+
"learning_rate": 9.636363636363637e-05,
|
| 684 |
+
"loss": 1.7171,
|
| 685 |
+
"step": 46000
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 1.4834351522498523,
|
| 689 |
+
"grad_norm": 56532.88671875,
|
| 690 |
+
"learning_rate": 9.631313131313132e-05,
|
| 691 |
+
"loss": 1.7146,
|
| 692 |
+
"step": 46500
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.4993858963520648,
|
| 696 |
+
"grad_norm": 53762.78515625,
|
| 697 |
+
"learning_rate": 9.626262626262627e-05,
|
| 698 |
+
"loss": 1.7107,
|
| 699 |
+
"step": 47000
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.5153366404542772,
|
| 703 |
+
"grad_norm": 54982.08203125,
|
| 704 |
+
"learning_rate": 9.621212121212123e-05,
|
| 705 |
+
"loss": 1.7085,
|
| 706 |
+
"step": 47500
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.5312873845564896,
|
| 710 |
+
"grad_norm": 57135.71484375,
|
| 711 |
+
"learning_rate": 9.616161616161616e-05,
|
| 712 |
+
"loss": 1.7063,
|
| 713 |
+
"step": 48000
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 1.547238128658702,
|
| 717 |
+
"grad_norm": 55469.52734375,
|
| 718 |
+
"learning_rate": 9.611111111111112e-05,
|
| 719 |
+
"loss": 1.7021,
|
| 720 |
+
"step": 48500
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 1.5631888727609144,
|
| 724 |
+
"grad_norm": 55527.140625,
|
| 725 |
+
"learning_rate": 9.606060606060606e-05,
|
| 726 |
+
"loss": 1.6959,
|
| 727 |
+
"step": 49000
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 1.5791396168631266,
|
| 731 |
+
"grad_norm": 55039.515625,
|
| 732 |
+
"learning_rate": 9.601010101010101e-05,
|
| 733 |
+
"loss": 1.6953,
|
| 734 |
+
"step": 49500
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 1.595090360965339,
|
| 738 |
+
"grad_norm": 58584.22265625,
|
| 739 |
+
"learning_rate": 9.595959595959596e-05,
|
| 740 |
+
"loss": 1.6916,
|
| 741 |
+
"step": 50000
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 1.595090360965339,
|
| 745 |
+
"eval_loss": 1.60930597782135,
|
| 746 |
+
"eval_runtime": 4918.9878,
|
| 747 |
+
"eval_samples_per_second": 203.919,
|
| 748 |
+
"eval_steps_per_second": 1.593,
|
| 749 |
+
"step": 50000
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 1.6110411050675513,
|
| 753 |
+
"grad_norm": 55009.78125,
|
| 754 |
+
"learning_rate": 9.590909090909092e-05,
|
| 755 |
+
"loss": 1.685,
|
| 756 |
+
"step": 50500
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 1.6269918491697637,
|
| 760 |
+
"grad_norm": 53817.55078125,
|
| 761 |
+
"learning_rate": 9.585858585858586e-05,
|
| 762 |
+
"loss": 1.684,
|
| 763 |
+
"step": 51000
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 1.6429425932719761,
|
| 767 |
+
"grad_norm": 53155.3359375,
|
| 768 |
+
"learning_rate": 9.580808080808081e-05,
|
| 769 |
+
"loss": 1.6813,
|
| 770 |
+
"step": 51500
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 1.6588933373741885,
|
| 774 |
+
"grad_norm": 55288.33984375,
|
| 775 |
+
"learning_rate": 9.575757575757576e-05,
|
| 776 |
+
"loss": 1.6776,
|
| 777 |
+
"step": 52000
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 1.674844081476401,
|
| 781 |
+
"grad_norm": 58538.25,
|
| 782 |
+
"learning_rate": 9.570707070707072e-05,
|
| 783 |
+
"loss": 1.6746,
|
| 784 |
+
"step": 52500
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 1.6907948255786134,
|
| 788 |
+
"grad_norm": 54227.25390625,
|
| 789 |
+
"learning_rate": 9.565656565656566e-05,
|
| 790 |
+
"loss": 1.6715,
|
| 791 |
+
"step": 53000
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 1.7067455696808256,
|
| 795 |
+
"grad_norm": 56011.28125,
|
| 796 |
+
"learning_rate": 9.560606060606061e-05,
|
| 797 |
+
"loss": 1.6701,
|
| 798 |
+
"step": 53500
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 1.722696313783038,
|
| 802 |
+
"grad_norm": 51203.140625,
|
| 803 |
+
"learning_rate": 9.555555555555557e-05,
|
| 804 |
+
"loss": 1.6695,
|
| 805 |
+
"step": 54000
|
| 806 |
+
},
|
| 807 |
+
{
|
| 808 |
+
"epoch": 1.7386470578852502,
|
| 809 |
+
"grad_norm": 53641.3203125,
|
| 810 |
+
"learning_rate": 9.550505050505051e-05,
|
| 811 |
+
"loss": 1.6613,
|
| 812 |
+
"step": 54500
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 1.7545978019874626,
|
| 816 |
+
"grad_norm": 55869.0234375,
|
| 817 |
+
"learning_rate": 9.545454545454546e-05,
|
| 818 |
+
"loss": 1.6598,
|
| 819 |
+
"step": 55000
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 1.770548546089675,
|
| 823 |
+
"grad_norm": 54982.5546875,
|
| 824 |
+
"learning_rate": 9.540404040404041e-05,
|
| 825 |
+
"loss": 1.6595,
|
| 826 |
+
"step": 55500
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 1.7864992901918875,
|
| 830 |
+
"grad_norm": 55537.86328125,
|
| 831 |
+
"learning_rate": 9.535353535353537e-05,
|
| 832 |
+
"loss": 1.6544,
|
| 833 |
+
"step": 56000
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 1.8024500342941,
|
| 837 |
+
"grad_norm": 52897.34375,
|
| 838 |
+
"learning_rate": 9.53030303030303e-05,
|
| 839 |
+
"loss": 1.6512,
|
| 840 |
+
"step": 56500
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 1.8184007783963123,
|
| 844 |
+
"grad_norm": 55306.94921875,
|
| 845 |
+
"learning_rate": 9.525252525252526e-05,
|
| 846 |
+
"loss": 1.6493,
|
| 847 |
+
"step": 57000
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 1.8343515224985245,
|
| 851 |
+
"grad_norm": 55051.29296875,
|
| 852 |
+
"learning_rate": 9.52020202020202e-05,
|
| 853 |
+
"loss": 1.6453,
|
| 854 |
+
"step": 57500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 1.850302266600737,
|
| 858 |
+
"grad_norm": 54118.21484375,
|
| 859 |
+
"learning_rate": 9.515151515151515e-05,
|
| 860 |
+
"loss": 1.6441,
|
| 861 |
+
"step": 58000
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 1.8662530107029491,
|
| 865 |
+
"grad_norm": 56088.76171875,
|
| 866 |
+
"learning_rate": 9.51010101010101e-05,
|
| 867 |
+
"loss": 1.6429,
|
| 868 |
+
"step": 58500
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 1.8822037548051616,
|
| 872 |
+
"grad_norm": 52600.8359375,
|
| 873 |
+
"learning_rate": 9.505050505050506e-05,
|
| 874 |
+
"loss": 1.636,
|
| 875 |
+
"step": 59000
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 1.898154498907374,
|
| 879 |
+
"grad_norm": 55023.9375,
|
| 880 |
+
"learning_rate": 9.5e-05,
|
| 881 |
+
"loss": 1.6333,
|
| 882 |
+
"step": 59500
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 1.9141052430095864,
|
| 886 |
+
"grad_norm": 54295.30859375,
|
| 887 |
+
"learning_rate": 9.494949494949495e-05,
|
| 888 |
+
"loss": 1.6335,
|
| 889 |
+
"step": 60000
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 1.9141052430095864,
|
| 893 |
+
"eval_loss": 1.5532509088516235,
|
| 894 |
+
"eval_runtime": 4919.5325,
|
| 895 |
+
"eval_samples_per_second": 203.896,
|
| 896 |
+
"eval_steps_per_second": 1.593,
|
| 897 |
+
"step": 60000
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.9300559871117988,
|
| 901 |
+
"grad_norm": 55173.84765625,
|
| 902 |
+
"learning_rate": 9.48989898989899e-05,
|
| 903 |
+
"loss": 1.63,
|
| 904 |
+
"step": 60500
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.9460067312140112,
|
| 908 |
+
"grad_norm": 53612.4921875,
|
| 909 |
+
"learning_rate": 9.484848484848486e-05,
|
| 910 |
+
"loss": 1.6286,
|
| 911 |
+
"step": 61000
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 1.9619574753162237,
|
| 915 |
+
"grad_norm": 55060.9140625,
|
| 916 |
+
"learning_rate": 9.47979797979798e-05,
|
| 917 |
+
"loss": 1.6263,
|
| 918 |
+
"step": 61500
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 1.9779082194184359,
|
| 922 |
+
"grad_norm": 54993.5625,
|
| 923 |
+
"learning_rate": 9.474747474747475e-05,
|
| 924 |
+
"loss": 1.6231,
|
| 925 |
+
"step": 62000
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 1.9938589635206483,
|
| 929 |
+
"grad_norm": 55949.11328125,
|
| 930 |
+
"learning_rate": 9.469696969696971e-05,
|
| 931 |
+
"loss": 1.6228,
|
| 932 |
+
"step": 62500
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 2.0098097076228605,
|
| 936 |
+
"grad_norm": 54883.53515625,
|
| 937 |
+
"learning_rate": 9.464646464646464e-05,
|
| 938 |
+
"loss": 1.6197,
|
| 939 |
+
"step": 63000
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 2.025760451725073,
|
| 943 |
+
"grad_norm": 57476.609375,
|
| 944 |
+
"learning_rate": 9.45959595959596e-05,
|
| 945 |
+
"loss": 1.6145,
|
| 946 |
+
"step": 63500
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 2.0417111958272853,
|
| 950 |
+
"grad_norm": 55516.81640625,
|
| 951 |
+
"learning_rate": 9.454545454545455e-05,
|
| 952 |
+
"loss": 1.6126,
|
| 953 |
+
"step": 64000
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 2.0576619399294978,
|
| 957 |
+
"grad_norm": 53755.71484375,
|
| 958 |
+
"learning_rate": 9.449494949494951e-05,
|
| 959 |
+
"loss": 1.6109,
|
| 960 |
+
"step": 64500
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 2.07361268403171,
|
| 964 |
+
"grad_norm": 54177.87890625,
|
| 965 |
+
"learning_rate": 9.444444444444444e-05,
|
| 966 |
+
"loss": 1.6105,
|
| 967 |
+
"step": 65000
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 2.0895634281339226,
|
| 971 |
+
"grad_norm": 56828.4765625,
|
| 972 |
+
"learning_rate": 9.43939393939394e-05,
|
| 973 |
+
"loss": 1.6075,
|
| 974 |
+
"step": 65500
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 2.105514172236135,
|
| 978 |
+
"grad_norm": 55464.76171875,
|
| 979 |
+
"learning_rate": 9.434343434343435e-05,
|
| 980 |
+
"loss": 1.606,
|
| 981 |
+
"step": 66000
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.121464916338347,
|
| 985 |
+
"grad_norm": 56928.1015625,
|
| 986 |
+
"learning_rate": 9.42929292929293e-05,
|
| 987 |
+
"loss": 1.6023,
|
| 988 |
+
"step": 66500
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 2.1374156604405594,
|
| 992 |
+
"grad_norm": 51041.0,
|
| 993 |
+
"learning_rate": 9.424242424242424e-05,
|
| 994 |
+
"loss": 1.6,
|
| 995 |
+
"step": 67000
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 2.153366404542772,
|
| 999 |
+
"grad_norm": 56379.828125,
|
| 1000 |
+
"learning_rate": 9.41919191919192e-05,
|
| 1001 |
+
"loss": 1.6003,
|
| 1002 |
+
"step": 67500
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 2.1693171486449843,
|
| 1006 |
+
"grad_norm": 55763.93359375,
|
| 1007 |
+
"learning_rate": 9.414141414141415e-05,
|
| 1008 |
+
"loss": 1.596,
|
| 1009 |
+
"step": 68000
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 2.1852678927471967,
|
| 1013 |
+
"grad_norm": 53785.359375,
|
| 1014 |
+
"learning_rate": 9.40909090909091e-05,
|
| 1015 |
+
"loss": 1.5932,
|
| 1016 |
+
"step": 68500
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 2.201218636849409,
|
| 1020 |
+
"grad_norm": 56003.0703125,
|
| 1021 |
+
"learning_rate": 9.404040404040404e-05,
|
| 1022 |
+
"loss": 1.5928,
|
| 1023 |
+
"step": 69000
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 2.2171693809516215,
|
| 1027 |
+
"grad_norm": 55811.99609375,
|
| 1028 |
+
"learning_rate": 9.3989898989899e-05,
|
| 1029 |
+
"loss": 1.5919,
|
| 1030 |
+
"step": 69500
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 2.233120125053834,
|
| 1034 |
+
"grad_norm": 56307.57421875,
|
| 1035 |
+
"learning_rate": 9.393939393939395e-05,
|
| 1036 |
+
"loss": 1.5869,
|
| 1037 |
+
"step": 70000
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 2.233120125053834,
|
| 1041 |
+
"eval_loss": 1.5101096630096436,
|
| 1042 |
+
"eval_runtime": 4917.8928,
|
| 1043 |
+
"eval_samples_per_second": 203.964,
|
| 1044 |
+
"eval_steps_per_second": 1.594,
|
| 1045 |
+
"step": 70000
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 2.249070869156046,
|
| 1049 |
+
"grad_norm": 53560.9296875,
|
| 1050 |
+
"learning_rate": 9.388888888888889e-05,
|
| 1051 |
+
"loss": 1.5883,
|
| 1052 |
+
"step": 70500
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 2.2650216132582583,
|
| 1056 |
+
"grad_norm": 56536.40625,
|
| 1057 |
+
"learning_rate": 9.383838383838385e-05,
|
| 1058 |
+
"loss": 1.585,
|
| 1059 |
+
"step": 71000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 2.2809723573604708,
|
| 1063 |
+
"grad_norm": 54454.96875,
|
| 1064 |
+
"learning_rate": 9.378787878787879e-05,
|
| 1065 |
+
"loss": 1.582,
|
| 1066 |
+
"step": 71500
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 2.296923101462683,
|
| 1070 |
+
"grad_norm": 55888.09375,
|
| 1071 |
+
"learning_rate": 9.373737373737375e-05,
|
| 1072 |
+
"loss": 1.5776,
|
| 1073 |
+
"step": 72000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 2.3128738455648956,
|
| 1077 |
+
"grad_norm": 55370.46484375,
|
| 1078 |
+
"learning_rate": 9.368686868686869e-05,
|
| 1079 |
+
"loss": 1.5781,
|
| 1080 |
+
"step": 72500
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 2.328824589667108,
|
| 1084 |
+
"grad_norm": 56668.328125,
|
| 1085 |
+
"learning_rate": 9.363636363636364e-05,
|
| 1086 |
+
"loss": 1.5779,
|
| 1087 |
+
"step": 73000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 2.3447753337693205,
|
| 1091 |
+
"grad_norm": 56674.85546875,
|
| 1092 |
+
"learning_rate": 9.358585858585858e-05,
|
| 1093 |
+
"loss": 1.5724,
|
| 1094 |
+
"step": 73500
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 2.360726077871533,
|
| 1098 |
+
"grad_norm": 59070.734375,
|
| 1099 |
+
"learning_rate": 9.353535353535354e-05,
|
| 1100 |
+
"loss": 1.5737,
|
| 1101 |
+
"step": 74000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 2.376676821973745,
|
| 1105 |
+
"grad_norm": 55701.37109375,
|
| 1106 |
+
"learning_rate": 9.348484848484849e-05,
|
| 1107 |
+
"loss": 1.5702,
|
| 1108 |
+
"step": 74500
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 2.3926275660759573,
|
| 1112 |
+
"grad_norm": 54837.890625,
|
| 1113 |
+
"learning_rate": 9.343434343434344e-05,
|
| 1114 |
+
"loss": 1.5691,
|
| 1115 |
+
"step": 75000
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 2.4085783101781697,
|
| 1119 |
+
"grad_norm": 55847.98046875,
|
| 1120 |
+
"learning_rate": 9.338383838383838e-05,
|
| 1121 |
+
"loss": 1.5692,
|
| 1122 |
+
"step": 75500
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 2.424529054280382,
|
| 1126 |
+
"grad_norm": 53633.03125,
|
| 1127 |
+
"learning_rate": 9.333333333333334e-05,
|
| 1128 |
+
"loss": 1.5639,
|
| 1129 |
+
"step": 76000
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 2.4404797983825945,
|
| 1133 |
+
"grad_norm": 55944.12890625,
|
| 1134 |
+
"learning_rate": 9.328282828282829e-05,
|
| 1135 |
+
"loss": 1.5664,
|
| 1136 |
+
"step": 76500
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 2.456430542484807,
|
| 1140 |
+
"grad_norm": 53979.30859375,
|
| 1141 |
+
"learning_rate": 9.323232323232324e-05,
|
| 1142 |
+
"loss": 1.5635,
|
| 1143 |
+
"step": 77000
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 2.4723812865870194,
|
| 1147 |
+
"grad_norm": 56014.97265625,
|
| 1148 |
+
"learning_rate": 9.318181818181818e-05,
|
| 1149 |
+
"loss": 1.5601,
|
| 1150 |
+
"step": 77500
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 2.488332030689232,
|
| 1154 |
+
"grad_norm": 55291.9140625,
|
| 1155 |
+
"learning_rate": 9.313131313131314e-05,
|
| 1156 |
+
"loss": 1.5618,
|
| 1157 |
+
"step": 78000
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 2.5042827747914442,
|
| 1161 |
+
"grad_norm": 53215.24609375,
|
| 1162 |
+
"learning_rate": 9.308080808080809e-05,
|
| 1163 |
+
"loss": 1.5576,
|
| 1164 |
+
"step": 78500
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 2.520233518893656,
|
| 1168 |
+
"grad_norm": 56197.3203125,
|
| 1169 |
+
"learning_rate": 9.303030303030303e-05,
|
| 1170 |
+
"loss": 1.554,
|
| 1171 |
+
"step": 79000
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 2.5361842629958686,
|
| 1175 |
+
"grad_norm": 55106.9765625,
|
| 1176 |
+
"learning_rate": 9.2979797979798e-05,
|
| 1177 |
+
"loss": 1.5545,
|
| 1178 |
+
"step": 79500
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 2.552135007098081,
|
| 1182 |
+
"grad_norm": 54552.08984375,
|
| 1183 |
+
"learning_rate": 9.292929292929293e-05,
|
| 1184 |
+
"loss": 1.5529,
|
| 1185 |
+
"step": 80000
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 2.552135007098081,
|
| 1189 |
+
"eval_loss": 1.4744161367416382,
|
| 1190 |
+
"eval_runtime": 4899.8032,
|
| 1191 |
+
"eval_samples_per_second": 204.717,
|
| 1192 |
+
"eval_steps_per_second": 1.599,
|
| 1193 |
+
"step": 80000
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 2.5681017019443955,
|
| 1197 |
+
"grad_norm": 56671.71875,
|
| 1198 |
+
"learning_rate": 9.287878787878789e-05,
|
| 1199 |
+
"loss": 1.5516,
|
| 1200 |
+
"step": 80500
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 2.5840524460466083,
|
| 1204 |
+
"grad_norm": 54676.5078125,
|
| 1205 |
+
"learning_rate": 9.282828282828283e-05,
|
| 1206 |
+
"loss": 1.5501,
|
| 1207 |
+
"step": 81000
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 2.6000031901488203,
|
| 1211 |
+
"grad_norm": 56587.88671875,
|
| 1212 |
+
"learning_rate": 9.277777777777778e-05,
|
| 1213 |
+
"loss": 1.549,
|
| 1214 |
+
"step": 81500
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 2.6159539342510327,
|
| 1218 |
+
"grad_norm": 56696.6484375,
|
| 1219 |
+
"learning_rate": 9.272727272727273e-05,
|
| 1220 |
+
"loss": 1.5448,
|
| 1221 |
+
"step": 82000
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 2.631904678353245,
|
| 1225 |
+
"grad_norm": 55145.359375,
|
| 1226 |
+
"learning_rate": 9.267676767676769e-05,
|
| 1227 |
+
"loss": 1.5439,
|
| 1228 |
+
"step": 82500
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 2.6478554224554576,
|
| 1232 |
+
"grad_norm": 55258.8828125,
|
| 1233 |
+
"learning_rate": 9.262626262626263e-05,
|
| 1234 |
+
"loss": 1.5461,
|
| 1235 |
+
"step": 83000
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 2.66380616655767,
|
| 1239 |
+
"grad_norm": 55989.4921875,
|
| 1240 |
+
"learning_rate": 9.257575757575758e-05,
|
| 1241 |
+
"loss": 1.5423,
|
| 1242 |
+
"step": 83500
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 2.6797569106598824,
|
| 1246 |
+
"grad_norm": 54569.66015625,
|
| 1247 |
+
"learning_rate": 9.252525252525253e-05,
|
| 1248 |
+
"loss": 1.5432,
|
| 1249 |
+
"step": 84000
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 2.695707654762095,
|
| 1253 |
+
"grad_norm": 56487.32421875,
|
| 1254 |
+
"learning_rate": 9.247474747474749e-05,
|
| 1255 |
+
"loss": 1.5406,
|
| 1256 |
+
"step": 84500
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 2.711658398864307,
|
| 1260 |
+
"grad_norm": 55735.38671875,
|
| 1261 |
+
"learning_rate": 9.242424242424242e-05,
|
| 1262 |
+
"loss": 1.5385,
|
| 1263 |
+
"step": 85000
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 2.7276091429665192,
|
| 1267 |
+
"grad_norm": 55246.734375,
|
| 1268 |
+
"learning_rate": 9.237373737373738e-05,
|
| 1269 |
+
"loss": 1.5385,
|
| 1270 |
+
"step": 85500
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 2.7435598870687317,
|
| 1274 |
+
"grad_norm": 54426.76171875,
|
| 1275 |
+
"learning_rate": 9.232323232323232e-05,
|
| 1276 |
+
"loss": 1.5354,
|
| 1277 |
+
"step": 86000
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 2.759510631170944,
|
| 1281 |
+
"grad_norm": 56496.05859375,
|
| 1282 |
+
"learning_rate": 9.227272727272727e-05,
|
| 1283 |
+
"loss": 1.5354,
|
| 1284 |
+
"step": 86500
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 2.7754613752731565,
|
| 1288 |
+
"grad_norm": 54483.66796875,
|
| 1289 |
+
"learning_rate": 9.222222222222223e-05,
|
| 1290 |
+
"loss": 1.5314,
|
| 1291 |
+
"step": 87000
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 2.791412119375369,
|
| 1295 |
+
"grad_norm": 56842.0,
|
| 1296 |
+
"learning_rate": 9.217171717171718e-05,
|
| 1297 |
+
"loss": 1.5307,
|
| 1298 |
+
"step": 87500
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 2.8073628634775813,
|
| 1302 |
+
"grad_norm": 55707.21875,
|
| 1303 |
+
"learning_rate": 9.212121212121214e-05,
|
| 1304 |
+
"loss": 1.5299,
|
| 1305 |
+
"step": 88000
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 2.8233136075797933,
|
| 1309 |
+
"grad_norm": 54965.58984375,
|
| 1310 |
+
"learning_rate": 9.207070707070707e-05,
|
| 1311 |
+
"loss": 1.5292,
|
| 1312 |
+
"step": 88500
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 2.839264351682006,
|
| 1316 |
+
"grad_norm": 54280.4140625,
|
| 1317 |
+
"learning_rate": 9.202020202020203e-05,
|
| 1318 |
+
"loss": 1.5281,
|
| 1319 |
+
"step": 89000
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 2.855215095784218,
|
| 1323 |
+
"grad_norm": 56317.328125,
|
| 1324 |
+
"learning_rate": 9.196969696969698e-05,
|
| 1325 |
+
"loss": 1.5292,
|
| 1326 |
+
"step": 89500
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 2.8711658398864306,
|
| 1330 |
+
"grad_norm": 56429.265625,
|
| 1331 |
+
"learning_rate": 9.191919191919192e-05,
|
| 1332 |
+
"loss": 1.5248,
|
| 1333 |
+
"step": 90000
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 2.8711658398864306,
|
| 1337 |
+
"eval_loss": 1.4486085176467896,
|
| 1338 |
+
"eval_runtime": 4949.3767,
|
| 1339 |
+
"eval_samples_per_second": 202.667,
|
| 1340 |
+
"eval_steps_per_second": 1.583,
|
| 1341 |
+
"step": 90000
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 2.887116583988643,
|
| 1345 |
+
"grad_norm": 60500.609375,
|
| 1346 |
+
"learning_rate": 9.186868686868687e-05,
|
| 1347 |
+
"loss": 1.5244,
|
| 1348 |
+
"step": 90500
|
| 1349 |
+
},
|
| 1350 |
+
{
|
| 1351 |
+
"epoch": 2.9030673280908554,
|
| 1352 |
+
"grad_norm": 56278.77734375,
|
| 1353 |
+
"learning_rate": 9.181818181818183e-05,
|
| 1354 |
+
"loss": 1.5228,
|
| 1355 |
+
"step": 91000
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 2.919018072193068,
|
| 1359 |
+
"grad_norm": 55179.0859375,
|
| 1360 |
+
"learning_rate": 9.176767676767677e-05,
|
| 1361 |
+
"loss": 1.5228,
|
| 1362 |
+
"step": 91500
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 2.9349688162952803,
|
| 1366 |
+
"grad_norm": 58535.69140625,
|
| 1367 |
+
"learning_rate": 9.171717171717172e-05,
|
| 1368 |
+
"loss": 1.5207,
|
| 1369 |
+
"step": 92000
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 2.9509195603974927,
|
| 1373 |
+
"grad_norm": 56315.96484375,
|
| 1374 |
+
"learning_rate": 9.166666666666667e-05,
|
| 1375 |
+
"loss": 1.5205,
|
| 1376 |
+
"step": 92500
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 2.9668703044997047,
|
| 1380 |
+
"grad_norm": 57173.87109375,
|
| 1381 |
+
"learning_rate": 9.161616161616163e-05,
|
| 1382 |
+
"loss": 1.5165,
|
| 1383 |
+
"step": 93000
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 2.9828210486019175,
|
| 1387 |
+
"grad_norm": 58435.5,
|
| 1388 |
+
"learning_rate": 9.156565656565656e-05,
|
| 1389 |
+
"loss": 1.516,
|
| 1390 |
+
"step": 93500
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 2.9987717927041295,
|
| 1394 |
+
"grad_norm": 55689.07421875,
|
| 1395 |
+
"learning_rate": 9.151515151515152e-05,
|
| 1396 |
+
"loss": 1.5155,
|
| 1397 |
+
"step": 94000
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 3.014722536806342,
|
| 1401 |
+
"grad_norm": 55691.28515625,
|
| 1402 |
+
"learning_rate": 9.146464646464647e-05,
|
| 1403 |
+
"loss": 1.5124,
|
| 1404 |
+
"step": 94500
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 3.0306732809085544,
|
| 1408 |
+
"grad_norm": 55951.5546875,
|
| 1409 |
+
"learning_rate": 9.141414141414141e-05,
|
| 1410 |
+
"loss": 1.5098,
|
| 1411 |
+
"step": 95000
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 3.046624025010767,
|
| 1415 |
+
"grad_norm": 56781.6328125,
|
| 1416 |
+
"learning_rate": 9.136363636363637e-05,
|
| 1417 |
+
"loss": 1.5107,
|
| 1418 |
+
"step": 95500
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 3.062574769112979,
|
| 1422 |
+
"grad_norm": 56111.234375,
|
| 1423 |
+
"learning_rate": 9.131313131313132e-05,
|
| 1424 |
+
"loss": 1.5072,
|
| 1425 |
+
"step": 96000
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 3.0785255132151916,
|
| 1429 |
+
"grad_norm": 55225.51171875,
|
| 1430 |
+
"learning_rate": 9.126262626262627e-05,
|
| 1431 |
+
"loss": 1.5089,
|
| 1432 |
+
"step": 96500
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"epoch": 3.094476257317404,
|
| 1436 |
+
"grad_norm": 57983.26171875,
|
| 1437 |
+
"learning_rate": 9.121212121212121e-05,
|
| 1438 |
+
"loss": 1.5068,
|
| 1439 |
+
"step": 97000
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 3.110427001419616,
|
| 1443 |
+
"grad_norm": 55611.953125,
|
| 1444 |
+
"learning_rate": 9.116161616161617e-05,
|
| 1445 |
+
"loss": 1.5039,
|
| 1446 |
+
"step": 97500
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 3.1263777455218285,
|
| 1450 |
+
"grad_norm": 56500.25390625,
|
| 1451 |
+
"learning_rate": 9.111111111111112e-05,
|
| 1452 |
+
"loss": 1.5023,
|
| 1453 |
+
"step": 98000
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 3.142328489624041,
|
| 1457 |
+
"grad_norm": 55070.7578125,
|
| 1458 |
+
"learning_rate": 9.106060606060606e-05,
|
| 1459 |
+
"loss": 1.5025,
|
| 1460 |
+
"step": 98500
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 3.1582792337262533,
|
| 1464 |
+
"grad_norm": 56306.3203125,
|
| 1465 |
+
"learning_rate": 9.101010101010101e-05,
|
| 1466 |
+
"loss": 1.501,
|
| 1467 |
+
"step": 99000
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 3.1742299778284657,
|
| 1471 |
+
"grad_norm": 56296.40234375,
|
| 1472 |
+
"learning_rate": 9.095959595959597e-05,
|
| 1473 |
+
"loss": 1.4999,
|
| 1474 |
+
"step": 99500
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 3.190180721930678,
|
| 1478 |
+
"grad_norm": 56137.7265625,
|
| 1479 |
+
"learning_rate": 9.090909090909092e-05,
|
| 1480 |
+
"loss": 1.5015,
|
| 1481 |
+
"step": 100000
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 3.190180721930678,
|
| 1485 |
+
"eval_loss": 1.4243189096450806,
|
| 1486 |
+
"eval_runtime": 4946.6072,
|
| 1487 |
+
"eval_samples_per_second": 202.78,
|
| 1488 |
+
"eval_steps_per_second": 1.584,
|
| 1489 |
+
"step": 100000
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 3.2061314660328906,
|
| 1493 |
+
"grad_norm": 57145.66796875,
|
| 1494 |
+
"learning_rate": 9.085858585858586e-05,
|
| 1495 |
+
"loss": 1.4976,
|
| 1496 |
+
"step": 100500
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 3.222082210135103,
|
| 1500 |
+
"grad_norm": 55889.25,
|
| 1501 |
+
"learning_rate": 9.080808080808081e-05,
|
| 1502 |
+
"loss": 1.4946,
|
| 1503 |
+
"step": 101000
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 3.238032954237315,
|
| 1507 |
+
"grad_norm": 54433.96875,
|
| 1508 |
+
"learning_rate": 9.075757575757577e-05,
|
| 1509 |
+
"loss": 1.4985,
|
| 1510 |
+
"step": 101500
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 3.2539836983395274,
|
| 1514 |
+
"grad_norm": 59956.953125,
|
| 1515 |
+
"learning_rate": 9.07070707070707e-05,
|
| 1516 |
+
"loss": 1.4926,
|
| 1517 |
+
"step": 102000
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 3.26993444244174,
|
| 1521 |
+
"grad_norm": 55148.5703125,
|
| 1522 |
+
"learning_rate": 9.065656565656566e-05,
|
| 1523 |
+
"loss": 1.4935,
|
| 1524 |
+
"step": 102500
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 3.2858851865439522,
|
| 1528 |
+
"grad_norm": 58131.62890625,
|
| 1529 |
+
"learning_rate": 9.060606060606061e-05,
|
| 1530 |
+
"loss": 1.4936,
|
| 1531 |
+
"step": 103000
|
| 1532 |
+
},
|
| 1533 |
+
{
|
| 1534 |
+
"epoch": 3.3018359306461647,
|
| 1535 |
+
"grad_norm": 61794.17578125,
|
| 1536 |
+
"learning_rate": 9.055555555555556e-05,
|
| 1537 |
+
"loss": 1.4922,
|
| 1538 |
+
"step": 103500
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 3.317786674748377,
|
| 1542 |
+
"grad_norm": 56916.46875,
|
| 1543 |
+
"learning_rate": 9.050505050505052e-05,
|
| 1544 |
+
"loss": 1.4915,
|
| 1545 |
+
"step": 104000
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 3.3337374188505895,
|
| 1549 |
+
"grad_norm": 56791.9765625,
|
| 1550 |
+
"learning_rate": 9.045454545454546e-05,
|
| 1551 |
+
"loss": 1.4885,
|
| 1552 |
+
"step": 104500
|
| 1553 |
+
},
|
| 1554 |
+
{
|
| 1555 |
+
"epoch": 3.349688162952802,
|
| 1556 |
+
"grad_norm": 59157.26171875,
|
| 1557 |
+
"learning_rate": 9.040404040404041e-05,
|
| 1558 |
+
"loss": 1.4892,
|
| 1559 |
+
"step": 105000
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 3.365638907055014,
|
| 1563 |
+
"grad_norm": 57222.37890625,
|
| 1564 |
+
"learning_rate": 9.035353535353535e-05,
|
| 1565 |
+
"loss": 1.4896,
|
| 1566 |
+
"step": 105500
|
| 1567 |
+
},
|
| 1568 |
+
{
|
| 1569 |
+
"epoch": 3.3815896511572263,
|
| 1570 |
+
"grad_norm": 59154.6171875,
|
| 1571 |
+
"learning_rate": 9.030303030303031e-05,
|
| 1572 |
+
"loss": 1.4846,
|
| 1573 |
+
"step": 106000
|
| 1574 |
+
},
|
| 1575 |
+
{
|
| 1576 |
+
"epoch": 3.3975403952594387,
|
| 1577 |
+
"grad_norm": 55996.48046875,
|
| 1578 |
+
"learning_rate": 9.025252525252526e-05,
|
| 1579 |
+
"loss": 1.4866,
|
| 1580 |
+
"step": 106500
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 3.413491139361651,
|
| 1584 |
+
"grad_norm": 58967.2578125,
|
| 1585 |
+
"learning_rate": 9.02020202020202e-05,
|
| 1586 |
+
"loss": 1.4846,
|
| 1587 |
+
"step": 107000
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 3.4294418834638636,
|
| 1591 |
+
"grad_norm": 58440.421875,
|
| 1592 |
+
"learning_rate": 9.015151515151515e-05,
|
| 1593 |
+
"loss": 1.4851,
|
| 1594 |
+
"step": 107500
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 3.445392627566076,
|
| 1598 |
+
"grad_norm": 55482.94140625,
|
| 1599 |
+
"learning_rate": 9.010101010101011e-05,
|
| 1600 |
+
"loss": 1.4862,
|
| 1601 |
+
"step": 108000
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 3.4613433716682884,
|
| 1605 |
+
"grad_norm": 59205.22265625,
|
| 1606 |
+
"learning_rate": 9.005050505050505e-05,
|
| 1607 |
+
"loss": 1.4829,
|
| 1608 |
+
"step": 108500
|
| 1609 |
+
},
|
| 1610 |
+
{
|
| 1611 |
+
"epoch": 3.477294115770501,
|
| 1612 |
+
"grad_norm": 54196.10546875,
|
| 1613 |
+
"learning_rate": 9e-05,
|
| 1614 |
+
"loss": 1.482,
|
| 1615 |
+
"step": 109000
|
| 1616 |
+
},
|
| 1617 |
+
{
|
| 1618 |
+
"epoch": 3.4932448598727133,
|
| 1619 |
+
"grad_norm": 54923.30078125,
|
| 1620 |
+
"learning_rate": 8.994949494949495e-05,
|
| 1621 |
+
"loss": 1.4797,
|
| 1622 |
+
"step": 109500
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 3.5091956039749252,
|
| 1626 |
+
"grad_norm": 54844.94140625,
|
| 1627 |
+
"learning_rate": 8.98989898989899e-05,
|
| 1628 |
+
"loss": 1.4788,
|
| 1629 |
+
"step": 110000
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 3.5091956039749252,
|
| 1633 |
+
"eval_loss": 1.4024385213851929,
|
| 1634 |
+
"eval_runtime": 4946.3399,
|
| 1635 |
+
"eval_samples_per_second": 202.791,
|
| 1636 |
+
"eval_steps_per_second": 1.584,
|
| 1637 |
+
"step": 110000
|
| 1638 |
+
}
|
| 1639 |
+
],
|
| 1640 |
+
"logging_steps": 500,
|
| 1641 |
+
"max_steps": 1000000,
|
| 1642 |
+
"num_input_tokens_seen": 0,
|
| 1643 |
+
"num_train_epochs": 32,
|
| 1644 |
+
"save_steps": 10000,
|
| 1645 |
+
"stateful_callbacks": {
|
| 1646 |
+
"TrainerControl": {
|
| 1647 |
+
"args": {
|
| 1648 |
+
"should_epoch_stop": false,
|
| 1649 |
+
"should_evaluate": false,
|
| 1650 |
+
"should_log": false,
|
| 1651 |
+
"should_save": true,
|
| 1652 |
+
"should_training_stop": false
|
| 1653 |
+
},
|
| 1654 |
+
"attributes": {}
|
| 1655 |
+
}
|
| 1656 |
+
},
|
| 1657 |
+
"total_flos": 7.413478046369151e+18,
|
| 1658 |
+
"train_batch_size": 128,
|
| 1659 |
+
"trial_name": null,
|
| 1660 |
+
"trial_params": null
|
| 1661 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0eadb0ed708f28746d449a9a1b7757971603512455947895bbaeed92d0c02a66
|
| 3 |
+
size 5304
|