Upload folder using huggingface_hub
Browse files- added_tokens.json +3 -0
- checkpoint-10656/config.json +28 -0
- checkpoint-10656/generation_config.json +7 -0
- checkpoint-10656/model.safetensors +3 -0
- checkpoint-10656/optimizer.pt +3 -0
- checkpoint-10656/rng_state.pth +3 -0
- checkpoint-10656/scheduler.pt +3 -0
- checkpoint-10656/trainer_state.json +3793 -0
- checkpoint-10656/training_args.bin +3 -0
- config.json +28 -0
- generation_config.json +7 -0
- merges.txt +0 -0
- model.safetensors +3 -0
- special_tokens_map.json +6 -0
- tokenizer.json +0 -0
- tokenizer_config.json +44 -0
- training_args.bin +3 -0
- vocab.json +0 -0
added_tokens.json
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"<|endoftext|>": 16000
|
3 |
+
}
|
checkpoint-10656/config.json
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"architectures": [
|
3 |
+
"LlamaForCausalLM"
|
4 |
+
],
|
5 |
+
"attention_bias": false,
|
6 |
+
"attention_dropout": 0.0,
|
7 |
+
"bos_token_id": 1,
|
8 |
+
"eos_token_id": 2,
|
9 |
+
"hidden_act": "silu",
|
10 |
+
"hidden_size": 512,
|
11 |
+
"initializer_range": 0.02,
|
12 |
+
"intermediate_size": 1024,
|
13 |
+
"max_position_embeddings": 256,
|
14 |
+
"model_type": "llama",
|
15 |
+
"num_attention_heads": 8,
|
16 |
+
"num_hidden_layers": 16,
|
17 |
+
"num_key_value_heads": 8,
|
18 |
+
"pad_token_id": 0,
|
19 |
+
"pretraining_tp": 1,
|
20 |
+
"rms_norm_eps": 1e-06,
|
21 |
+
"rope_scaling": null,
|
22 |
+
"rope_theta": 10000.0,
|
23 |
+
"tie_word_embeddings": false,
|
24 |
+
"torch_dtype": "float32",
|
25 |
+
"transformers_version": "4.40.0",
|
26 |
+
"use_cache": true,
|
27 |
+
"vocab_size": 16000
|
28 |
+
}
|
checkpoint-10656/generation_config.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"bos_token_id": 1,
|
4 |
+
"eos_token_id": 2,
|
5 |
+
"pad_token_id": 0,
|
6 |
+
"transformers_version": "4.40.0"
|
7 |
+
}
|
checkpoint-10656/model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:466f10e3a8cb52f000f8c00350a8c6cd3b33199c74ee996676395ea8d940e2e8
|
3 |
+
size 233392032
|
checkpoint-10656/optimizer.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5ce7edfdf9270445de4ef2fc49a0f4e2d27dfd27fdb6737522bbbc3c0c859730
|
3 |
+
size 466875514
|
checkpoint-10656/rng_state.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a83b2296c01ff59b01a91dcca81f8f5808e35d9710aa393f6be21cb85b7282f9
|
3 |
+
size 14244
|
checkpoint-10656/scheduler.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2ce515755abaae233c0fb44ae0343ebd5435a3729c7694ac5580f797adeec224
|
3 |
+
size 1064
|
checkpoint-10656/trainer_state.json
ADDED
@@ -0,0 +1,3793 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 5.124037742614746,
|
3 |
+
"best_model_checkpoint": "/home/ubuntu/amazon_capstone/code/SJK/BabyLlama/models/Baby-Llama-58M_Orig/checkpoint-10656",
|
4 |
+
"epoch": 6.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 10656,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.01126126126126126,
|
13 |
+
"grad_norm": 103.2569351196289,
|
14 |
+
"learning_rate": 2.2499999999999998e-05,
|
15 |
+
"loss": 267.2774,
|
16 |
+
"step": 20
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.02252252252252252,
|
20 |
+
"grad_norm": 56.244712829589844,
|
21 |
+
"learning_rate": 4.625e-05,
|
22 |
+
"loss": 250.6269,
|
23 |
+
"step": 40
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.033783783783783786,
|
27 |
+
"grad_norm": 60.72817611694336,
|
28 |
+
"learning_rate": 7.125e-05,
|
29 |
+
"loss": 223.0035,
|
30 |
+
"step": 60
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.04504504504504504,
|
34 |
+
"grad_norm": 57.53702163696289,
|
35 |
+
"learning_rate": 9.625000000000001e-05,
|
36 |
+
"loss": 187.8246,
|
37 |
+
"step": 80
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.05630630630630631,
|
41 |
+
"grad_norm": 52.90182113647461,
|
42 |
+
"learning_rate": 0.00012124999999999999,
|
43 |
+
"loss": 164.3489,
|
44 |
+
"step": 100
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.06756756756756757,
|
48 |
+
"grad_norm": 41.146644592285156,
|
49 |
+
"learning_rate": 0.00014625,
|
50 |
+
"loss": 140.3743,
|
51 |
+
"step": 120
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.07882882882882883,
|
55 |
+
"grad_norm": 51.88545608520508,
|
56 |
+
"learning_rate": 0.00017125000000000002,
|
57 |
+
"loss": 127.1389,
|
58 |
+
"step": 140
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.09009009009009009,
|
62 |
+
"grad_norm": 46.53668975830078,
|
63 |
+
"learning_rate": 0.00019625,
|
64 |
+
"loss": 112.8051,
|
65 |
+
"step": 160
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.10135135135135136,
|
69 |
+
"grad_norm": 39.3553352355957,
|
70 |
+
"learning_rate": 0.00022125,
|
71 |
+
"loss": 104.9926,
|
72 |
+
"step": 180
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.11261261261261261,
|
76 |
+
"grad_norm": 58.92243194580078,
|
77 |
+
"learning_rate": 0.00024625,
|
78 |
+
"loss": 93.4464,
|
79 |
+
"step": 200
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.12387387387387387,
|
83 |
+
"grad_norm": 55.10353088378906,
|
84 |
+
"learning_rate": 0.0002499983694072308,
|
85 |
+
"loss": 86.6169,
|
86 |
+
"step": 220
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.13513513513513514,
|
90 |
+
"grad_norm": 35.936737060546875,
|
91 |
+
"learning_rate": 0.0002499922759053138,
|
92 |
+
"loss": 77.5335,
|
93 |
+
"step": 240
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.1463963963963964,
|
97 |
+
"grad_norm": 36.4266357421875,
|
98 |
+
"learning_rate": 0.0002499816689344726,
|
99 |
+
"loss": 71.1157,
|
100 |
+
"step": 260
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.15765765765765766,
|
104 |
+
"grad_norm": 53.89732360839844,
|
105 |
+
"learning_rate": 0.00024996654887772486,
|
106 |
+
"loss": 64.3968,
|
107 |
+
"step": 280
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.16891891891891891,
|
111 |
+
"grad_norm": 34.17560958862305,
|
112 |
+
"learning_rate": 0.0002499469162810555,
|
113 |
+
"loss": 63.59,
|
114 |
+
"step": 300
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.18018018018018017,
|
118 |
+
"grad_norm": 36.95146179199219,
|
119 |
+
"learning_rate": 0.0002499227718533972,
|
120 |
+
"loss": 57.8954,
|
121 |
+
"step": 320
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.19144144144144143,
|
125 |
+
"grad_norm": 27.419086456298828,
|
126 |
+
"learning_rate": 0.0002498941164666049,
|
127 |
+
"loss": 55.4757,
|
128 |
+
"step": 340
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.20270270270270271,
|
132 |
+
"grad_norm": 44.045127868652344,
|
133 |
+
"learning_rate": 0.0002498609511554241,
|
134 |
+
"loss": 52.6704,
|
135 |
+
"step": 360
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.21396396396396397,
|
139 |
+
"grad_norm": 42.809051513671875,
|
140 |
+
"learning_rate": 0.00024982327711745364,
|
141 |
+
"loss": 51.5728,
|
142 |
+
"step": 380
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.22522522522522523,
|
146 |
+
"grad_norm": 37.47595977783203,
|
147 |
+
"learning_rate": 0.00024978109571310224,
|
148 |
+
"loss": 48.3088,
|
149 |
+
"step": 400
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.23648648648648649,
|
153 |
+
"grad_norm": 32.29904556274414,
|
154 |
+
"learning_rate": 0.0002497344084655399,
|
155 |
+
"loss": 46.0072,
|
156 |
+
"step": 420
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 0.24774774774774774,
|
160 |
+
"grad_norm": 33.23353576660156,
|
161 |
+
"learning_rate": 0.0002496832170606422,
|
162 |
+
"loss": 44.5736,
|
163 |
+
"step": 440
|
164 |
+
},
|
165 |
+
{
|
166 |
+
"epoch": 0.25900900900900903,
|
167 |
+
"grad_norm": 27.061250686645508,
|
168 |
+
"learning_rate": 0.00024962752334692985,
|
169 |
+
"loss": 40.7675,
|
170 |
+
"step": 460
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.2702702702702703,
|
174 |
+
"grad_norm": 38.74943161010742,
|
175 |
+
"learning_rate": 0.000249567329335502,
|
176 |
+
"loss": 41.9667,
|
177 |
+
"step": 480
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.28153153153153154,
|
181 |
+
"grad_norm": 25.483449935913086,
|
182 |
+
"learning_rate": 0.0002495026371999633,
|
183 |
+
"loss": 39.6405,
|
184 |
+
"step": 500
|
185 |
+
},
|
186 |
+
{
|
187 |
+
"epoch": 0.2927927927927928,
|
188 |
+
"grad_norm": 31.223539352416992,
|
189 |
+
"learning_rate": 0.0002494334492763456,
|
190 |
+
"loss": 37.7263,
|
191 |
+
"step": 520
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 0.30405405405405406,
|
195 |
+
"grad_norm": 28.23072624206543,
|
196 |
+
"learning_rate": 0.0002493597680630237,
|
197 |
+
"loss": 35.0503,
|
198 |
+
"step": 540
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 0.3153153153153153,
|
202 |
+
"grad_norm": 26.49002456665039,
|
203 |
+
"learning_rate": 0.00024928159622062497,
|
204 |
+
"loss": 35.0304,
|
205 |
+
"step": 560
|
206 |
+
},
|
207 |
+
{
|
208 |
+
"epoch": 0.32657657657657657,
|
209 |
+
"grad_norm": 33.025638580322266,
|
210 |
+
"learning_rate": 0.0002491989365719332,
|
211 |
+
"loss": 33.0983,
|
212 |
+
"step": 580
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.33783783783783783,
|
216 |
+
"grad_norm": 22.15628433227539,
|
217 |
+
"learning_rate": 0.00024911179210178693,
|
218 |
+
"loss": 31.852,
|
219 |
+
"step": 600
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.3490990990990991,
|
223 |
+
"grad_norm": 24.4526424407959,
|
224 |
+
"learning_rate": 0.0002490201659569714,
|
225 |
+
"loss": 30.807,
|
226 |
+
"step": 620
|
227 |
+
},
|
228 |
+
{
|
229 |
+
"epoch": 0.36036036036036034,
|
230 |
+
"grad_norm": 29.9416561126709,
|
231 |
+
"learning_rate": 0.00024892406144610523,
|
232 |
+
"loss": 30.2397,
|
233 |
+
"step": 640
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 0.3716216216216216,
|
237 |
+
"grad_norm": 23.99987030029297,
|
238 |
+
"learning_rate": 0.0002488234820395206,
|
239 |
+
"loss": 29.0914,
|
240 |
+
"step": 660
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 0.38288288288288286,
|
244 |
+
"grad_norm": 26.059839248657227,
|
245 |
+
"learning_rate": 0.0002487184313691381,
|
246 |
+
"loss": 28.032,
|
247 |
+
"step": 680
|
248 |
+
},
|
249 |
+
{
|
250 |
+
"epoch": 0.39414414414414417,
|
251 |
+
"grad_norm": 21.169851303100586,
|
252 |
+
"learning_rate": 0.00024860891322833566,
|
253 |
+
"loss": 26.5924,
|
254 |
+
"step": 700
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.40540540540540543,
|
258 |
+
"grad_norm": 24.150981903076172,
|
259 |
+
"learning_rate": 0.0002484949315718115,
|
260 |
+
"loss": 26.607,
|
261 |
+
"step": 720
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 0.4166666666666667,
|
265 |
+
"grad_norm": 23.67961883544922,
|
266 |
+
"learning_rate": 0.00024837649051544117,
|
267 |
+
"loss": 26.0378,
|
268 |
+
"step": 740
|
269 |
+
},
|
270 |
+
{
|
271 |
+
"epoch": 0.42792792792792794,
|
272 |
+
"grad_norm": 22.1843204498291,
|
273 |
+
"learning_rate": 0.00024825359433612923,
|
274 |
+
"loss": 24.8562,
|
275 |
+
"step": 760
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 0.4391891891891892,
|
279 |
+
"grad_norm": 24.654254913330078,
|
280 |
+
"learning_rate": 0.0002481262474716546,
|
281 |
+
"loss": 24.8206,
|
282 |
+
"step": 780
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 0.45045045045045046,
|
286 |
+
"grad_norm": 22.423748016357422,
|
287 |
+
"learning_rate": 0.00024799445452051034,
|
288 |
+
"loss": 23.547,
|
289 |
+
"step": 800
|
290 |
+
},
|
291 |
+
{
|
292 |
+
"epoch": 0.4617117117117117,
|
293 |
+
"grad_norm": 21.533939361572266,
|
294 |
+
"learning_rate": 0.0002478582202417376,
|
295 |
+
"loss": 22.8409,
|
296 |
+
"step": 820
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.47297297297297297,
|
300 |
+
"grad_norm": 22.048704147338867,
|
301 |
+
"learning_rate": 0.0002477175495547539,
|
302 |
+
"loss": 22.1961,
|
303 |
+
"step": 840
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 0.48423423423423423,
|
307 |
+
"grad_norm": 30.49504852294922,
|
308 |
+
"learning_rate": 0.0002475724475391753,
|
309 |
+
"loss": 21.7862,
|
310 |
+
"step": 860
|
311 |
+
},
|
312 |
+
{
|
313 |
+
"epoch": 0.4954954954954955,
|
314 |
+
"grad_norm": 21.017244338989258,
|
315 |
+
"learning_rate": 0.0002474229194346331,
|
316 |
+
"loss": 20.896,
|
317 |
+
"step": 880
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 0.5067567567567568,
|
321 |
+
"grad_norm": 18.561767578125,
|
322 |
+
"learning_rate": 0.00024726897064058443,
|
323 |
+
"loss": 20.7818,
|
324 |
+
"step": 900
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 0.5180180180180181,
|
328 |
+
"grad_norm": 21.154193878173828,
|
329 |
+
"learning_rate": 0.0002471106067161178,
|
330 |
+
"loss": 20.2752,
|
331 |
+
"step": 920
|
332 |
+
},
|
333 |
+
{
|
334 |
+
"epoch": 0.5292792792792793,
|
335 |
+
"grad_norm": 17.5256290435791,
|
336 |
+
"learning_rate": 0.0002469478333797516,
|
337 |
+
"loss": 19.5015,
|
338 |
+
"step": 940
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.5405405405405406,
|
342 |
+
"grad_norm": 21.720947265625,
|
343 |
+
"learning_rate": 0.0002467806565092284,
|
344 |
+
"loss": 19.4266,
|
345 |
+
"step": 960
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 0.5518018018018018,
|
349 |
+
"grad_norm": 23.965906143188477,
|
350 |
+
"learning_rate": 0.00024660908214130185,
|
351 |
+
"loss": 18.5526,
|
352 |
+
"step": 980
|
353 |
+
},
|
354 |
+
{
|
355 |
+
"epoch": 0.5630630630630631,
|
356 |
+
"grad_norm": 18.93579864501953,
|
357 |
+
"learning_rate": 0.0002464331164715196,
|
358 |
+
"loss": 17.9695,
|
359 |
+
"step": 1000
|
360 |
+
},
|
361 |
+
{
|
362 |
+
"epoch": 0.5743243243243243,
|
363 |
+
"grad_norm": 15.295672416687012,
|
364 |
+
"learning_rate": 0.00024625276585399884,
|
365 |
+
"loss": 17.5248,
|
366 |
+
"step": 1020
|
367 |
+
},
|
368 |
+
{
|
369 |
+
"epoch": 0.5855855855855856,
|
370 |
+
"grad_norm": 15.176102638244629,
|
371 |
+
"learning_rate": 0.0002460680368011974,
|
372 |
+
"loss": 17.407,
|
373 |
+
"step": 1040
|
374 |
+
},
|
375 |
+
{
|
376 |
+
"epoch": 0.5968468468468469,
|
377 |
+
"grad_norm": 16.598886489868164,
|
378 |
+
"learning_rate": 0.00024587893598367803,
|
379 |
+
"loss": 17.3692,
|
380 |
+
"step": 1060
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 0.6081081081081081,
|
384 |
+
"grad_norm": 17.836767196655273,
|
385 |
+
"learning_rate": 0.00024568547022986805,
|
386 |
+
"loss": 16.7794,
|
387 |
+
"step": 1080
|
388 |
+
},
|
389 |
+
{
|
390 |
+
"epoch": 0.6193693693693694,
|
391 |
+
"grad_norm": 17.016878128051758,
|
392 |
+
"learning_rate": 0.0002454876465258125,
|
393 |
+
"loss": 16.0592,
|
394 |
+
"step": 1100
|
395 |
+
},
|
396 |
+
{
|
397 |
+
"epoch": 0.6306306306306306,
|
398 |
+
"grad_norm": 18.94131088256836,
|
399 |
+
"learning_rate": 0.0002452854720149219,
|
400 |
+
"loss": 15.9458,
|
401 |
+
"step": 1120
|
402 |
+
},
|
403 |
+
{
|
404 |
+
"epoch": 0.6418918918918919,
|
405 |
+
"grad_norm": 20.51512908935547,
|
406 |
+
"learning_rate": 0.00024507895399771435,
|
407 |
+
"loss": 15.7893,
|
408 |
+
"step": 1140
|
409 |
+
},
|
410 |
+
{
|
411 |
+
"epoch": 0.6531531531531531,
|
412 |
+
"grad_norm": 18.29818344116211,
|
413 |
+
"learning_rate": 0.00024486809993155183,
|
414 |
+
"loss": 15.68,
|
415 |
+
"step": 1160
|
416 |
+
},
|
417 |
+
{
|
418 |
+
"epoch": 0.6644144144144144,
|
419 |
+
"grad_norm": 16.939292907714844,
|
420 |
+
"learning_rate": 0.00024465291743037114,
|
421 |
+
"loss": 15.8475,
|
422 |
+
"step": 1180
|
423 |
+
},
|
424 |
+
{
|
425 |
+
"epoch": 0.6756756756756757,
|
426 |
+
"grad_norm": 26.32220458984375,
|
427 |
+
"learning_rate": 0.00024443341426440865,
|
428 |
+
"loss": 14.9604,
|
429 |
+
"step": 1200
|
430 |
+
},
|
431 |
+
{
|
432 |
+
"epoch": 0.6869369369369369,
|
433 |
+
"grad_norm": 16.843460083007812,
|
434 |
+
"learning_rate": 0.0002442095983599198,
|
435 |
+
"loss": 15.0428,
|
436 |
+
"step": 1220
|
437 |
+
},
|
438 |
+
{
|
439 |
+
"epoch": 0.6981981981981982,
|
440 |
+
"grad_norm": 17.47309684753418,
|
441 |
+
"learning_rate": 0.0002439814777988931,
|
442 |
+
"loss": 14.6766,
|
443 |
+
"step": 1240
|
444 |
+
},
|
445 |
+
{
|
446 |
+
"epoch": 0.7094594594594594,
|
447 |
+
"grad_norm": 14.182714462280273,
|
448 |
+
"learning_rate": 0.00024374906081875807,
|
449 |
+
"loss": 14.3348,
|
450 |
+
"step": 1260
|
451 |
+
},
|
452 |
+
{
|
453 |
+
"epoch": 0.7207207207207207,
|
454 |
+
"grad_norm": 14.923556327819824,
|
455 |
+
"learning_rate": 0.00024351235581208786,
|
456 |
+
"loss": 14.1211,
|
457 |
+
"step": 1280
|
458 |
+
},
|
459 |
+
{
|
460 |
+
"epoch": 0.7319819819819819,
|
461 |
+
"grad_norm": 16.381290435791016,
|
462 |
+
"learning_rate": 0.00024327137132629614,
|
463 |
+
"loss": 13.6878,
|
464 |
+
"step": 1300
|
465 |
+
},
|
466 |
+
{
|
467 |
+
"epoch": 0.7432432432432432,
|
468 |
+
"grad_norm": 12.33255672454834,
|
469 |
+
"learning_rate": 0.00024302611606332857,
|
470 |
+
"loss": 13.8257,
|
471 |
+
"step": 1320
|
472 |
+
},
|
473 |
+
{
|
474 |
+
"epoch": 0.7545045045045045,
|
475 |
+
"grad_norm": 12.836396217346191,
|
476 |
+
"learning_rate": 0.00024277659887934845,
|
477 |
+
"loss": 13.2505,
|
478 |
+
"step": 1340
|
479 |
+
},
|
480 |
+
{
|
481 |
+
"epoch": 0.7657657657657657,
|
482 |
+
"grad_norm": 13.514806747436523,
|
483 |
+
"learning_rate": 0.00024252282878441694,
|
484 |
+
"loss": 12.9545,
|
485 |
+
"step": 1360
|
486 |
+
},
|
487 |
+
{
|
488 |
+
"epoch": 0.777027027027027,
|
489 |
+
"grad_norm": 13.225746154785156,
|
490 |
+
"learning_rate": 0.00024226481494216782,
|
491 |
+
"loss": 13.11,
|
492 |
+
"step": 1380
|
493 |
+
},
|
494 |
+
{
|
495 |
+
"epoch": 0.7882882882882883,
|
496 |
+
"grad_norm": 12.615843772888184,
|
497 |
+
"learning_rate": 0.00024200256666947648,
|
498 |
+
"loss": 12.9861,
|
499 |
+
"step": 1400
|
500 |
+
},
|
501 |
+
{
|
502 |
+
"epoch": 0.7995495495495496,
|
503 |
+
"grad_norm": 11.677081108093262,
|
504 |
+
"learning_rate": 0.00024173609343612352,
|
505 |
+
"loss": 13.1024,
|
506 |
+
"step": 1420
|
507 |
+
},
|
508 |
+
{
|
509 |
+
"epoch": 0.8108108108108109,
|
510 |
+
"grad_norm": 12.57310962677002,
|
511 |
+
"learning_rate": 0.0002414654048644528,
|
512 |
+
"loss": 12.8446,
|
513 |
+
"step": 1440
|
514 |
+
},
|
515 |
+
{
|
516 |
+
"epoch": 0.8220720720720721,
|
517 |
+
"grad_norm": 12.706310272216797,
|
518 |
+
"learning_rate": 0.0002411905107290239,
|
519 |
+
"loss": 12.6714,
|
520 |
+
"step": 1460
|
521 |
+
},
|
522 |
+
{
|
523 |
+
"epoch": 0.8333333333333334,
|
524 |
+
"grad_norm": 12.55820369720459,
|
525 |
+
"learning_rate": 0.0002409114209562594,
|
526 |
+
"loss": 12.2702,
|
527 |
+
"step": 1480
|
528 |
+
},
|
529 |
+
{
|
530 |
+
"epoch": 0.8445945945945946,
|
531 |
+
"grad_norm": 10.394694328308105,
|
532 |
+
"learning_rate": 0.0002406281456240861,
|
533 |
+
"loss": 11.8798,
|
534 |
+
"step": 1500
|
535 |
+
},
|
536 |
+
{
|
537 |
+
"epoch": 0.8558558558558559,
|
538 |
+
"grad_norm": 13.945009231567383,
|
539 |
+
"learning_rate": 0.0002403406949615714,
|
540 |
+
"loss": 11.9807,
|
541 |
+
"step": 1520
|
542 |
+
},
|
543 |
+
{
|
544 |
+
"epoch": 0.8671171171171171,
|
545 |
+
"grad_norm": 10.771109580993652,
|
546 |
+
"learning_rate": 0.00024004907934855373,
|
547 |
+
"loss": 11.7541,
|
548 |
+
"step": 1540
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 0.8783783783783784,
|
552 |
+
"grad_norm": 13.699746131896973,
|
553 |
+
"learning_rate": 0.00023975330931526784,
|
554 |
+
"loss": 11.8184,
|
555 |
+
"step": 1560
|
556 |
+
},
|
557 |
+
{
|
558 |
+
"epoch": 0.8896396396396397,
|
559 |
+
"grad_norm": 11.741015434265137,
|
560 |
+
"learning_rate": 0.00023945339554196456,
|
561 |
+
"loss": 11.583,
|
562 |
+
"step": 1580
|
563 |
+
},
|
564 |
+
{
|
565 |
+
"epoch": 0.9009009009009009,
|
566 |
+
"grad_norm": 12.183013916015625,
|
567 |
+
"learning_rate": 0.00023914934885852505,
|
568 |
+
"loss": 11.3534,
|
569 |
+
"step": 1600
|
570 |
+
},
|
571 |
+
{
|
572 |
+
"epoch": 0.9121621621621622,
|
573 |
+
"grad_norm": 11.609010696411133,
|
574 |
+
"learning_rate": 0.00023884118024406975,
|
575 |
+
"loss": 11.418,
|
576 |
+
"step": 1620
|
577 |
+
},
|
578 |
+
{
|
579 |
+
"epoch": 0.9234234234234234,
|
580 |
+
"grad_norm": 11.524951934814453,
|
581 |
+
"learning_rate": 0.00023852890082656203,
|
582 |
+
"loss": 11.3926,
|
583 |
+
"step": 1640
|
584 |
+
},
|
585 |
+
{
|
586 |
+
"epoch": 0.9346846846846847,
|
587 |
+
"grad_norm": 11.058307647705078,
|
588 |
+
"learning_rate": 0.00023821252188240634,
|
589 |
+
"loss": 11.0091,
|
590 |
+
"step": 1660
|
591 |
+
},
|
592 |
+
{
|
593 |
+
"epoch": 0.9459459459459459,
|
594 |
+
"grad_norm": 8.365026473999023,
|
595 |
+
"learning_rate": 0.00023789205483604076,
|
596 |
+
"loss": 10.6156,
|
597 |
+
"step": 1680
|
598 |
+
},
|
599 |
+
{
|
600 |
+
"epoch": 0.9572072072072072,
|
601 |
+
"grad_norm": 9.224916458129883,
|
602 |
+
"learning_rate": 0.0002375675112595249,
|
603 |
+
"loss": 11.1918,
|
604 |
+
"step": 1700
|
605 |
+
},
|
606 |
+
{
|
607 |
+
"epoch": 0.9684684684684685,
|
608 |
+
"grad_norm": 12.44520092010498,
|
609 |
+
"learning_rate": 0.00023723890287212162,
|
610 |
+
"loss": 10.8307,
|
611 |
+
"step": 1720
|
612 |
+
},
|
613 |
+
{
|
614 |
+
"epoch": 0.9797297297297297,
|
615 |
+
"grad_norm": 12.0541410446167,
|
616 |
+
"learning_rate": 0.00023690624153987423,
|
617 |
+
"loss": 10.7466,
|
618 |
+
"step": 1740
|
619 |
+
},
|
620 |
+
{
|
621 |
+
"epoch": 0.990990990990991,
|
622 |
+
"grad_norm": 11.42975902557373,
|
623 |
+
"learning_rate": 0.00023656953927517755,
|
624 |
+
"loss": 10.3984,
|
625 |
+
"step": 1760
|
626 |
+
},
|
627 |
+
{
|
628 |
+
"epoch": 1.0,
|
629 |
+
"eval_loss": 10.209589004516602,
|
630 |
+
"eval_runtime": 192.8583,
|
631 |
+
"eval_samples_per_second": 42.477,
|
632 |
+
"eval_steps_per_second": 5.31,
|
633 |
+
"step": 1776
|
634 |
+
},
|
635 |
+
{
|
636 |
+
"epoch": 1.0022522522522523,
|
637 |
+
"grad_norm": 10.396323204040527,
|
638 |
+
"learning_rate": 0.00023622880823634463,
|
639 |
+
"loss": 10.6677,
|
640 |
+
"step": 1780
|
641 |
+
},
|
642 |
+
{
|
643 |
+
"epoch": 1.0135135135135136,
|
644 |
+
"grad_norm": 9.66658878326416,
|
645 |
+
"learning_rate": 0.0002358840607271674,
|
646 |
+
"loss": 10.3207,
|
647 |
+
"step": 1800
|
648 |
+
},
|
649 |
+
{
|
650 |
+
"epoch": 1.0247747747747749,
|
651 |
+
"grad_norm": 9.684843063354492,
|
652 |
+
"learning_rate": 0.00023553530919647242,
|
653 |
+
"loss": 10.2381,
|
654 |
+
"step": 1820
|
655 |
+
},
|
656 |
+
{
|
657 |
+
"epoch": 1.0360360360360361,
|
658 |
+
"grad_norm": 8.732488632202148,
|
659 |
+
"learning_rate": 0.00023518256623767148,
|
660 |
+
"loss": 10.1474,
|
661 |
+
"step": 1840
|
662 |
+
},
|
663 |
+
{
|
664 |
+
"epoch": 1.0472972972972974,
|
665 |
+
"grad_norm": 9.16722297668457,
|
666 |
+
"learning_rate": 0.00023482584458830668,
|
667 |
+
"loss": 10.004,
|
668 |
+
"step": 1860
|
669 |
+
},
|
670 |
+
{
|
671 |
+
"epoch": 1.0585585585585586,
|
672 |
+
"grad_norm": 9.978281021118164,
|
673 |
+
"learning_rate": 0.00023446515712959072,
|
674 |
+
"loss": 9.9609,
|
675 |
+
"step": 1880
|
676 |
+
},
|
677 |
+
{
|
678 |
+
"epoch": 1.0698198198198199,
|
679 |
+
"grad_norm": 8.478277206420898,
|
680 |
+
"learning_rate": 0.00023410051688594135,
|
681 |
+
"loss": 10.1065,
|
682 |
+
"step": 1900
|
683 |
+
},
|
684 |
+
{
|
685 |
+
"epoch": 1.0810810810810811,
|
686 |
+
"grad_norm": 9.539369583129883,
|
687 |
+
"learning_rate": 0.00023373193702451163,
|
688 |
+
"loss": 9.8355,
|
689 |
+
"step": 1920
|
690 |
+
},
|
691 |
+
{
|
692 |
+
"epoch": 1.0923423423423424,
|
693 |
+
"grad_norm": 9.152040481567383,
|
694 |
+
"learning_rate": 0.00023335943085471387,
|
695 |
+
"loss": 9.6821,
|
696 |
+
"step": 1940
|
697 |
+
},
|
698 |
+
{
|
699 |
+
"epoch": 1.1036036036036037,
|
700 |
+
"grad_norm": 10.809807777404785,
|
701 |
+
"learning_rate": 0.00023298301182773947,
|
702 |
+
"loss": 9.8334,
|
703 |
+
"step": 1960
|
704 |
+
},
|
705 |
+
{
|
706 |
+
"epoch": 1.114864864864865,
|
707 |
+
"grad_norm": 8.823080062866211,
|
708 |
+
"learning_rate": 0.00023260269353607302,
|
709 |
+
"loss": 9.5618,
|
710 |
+
"step": 1980
|
711 |
+
},
|
712 |
+
{
|
713 |
+
"epoch": 1.1261261261261262,
|
714 |
+
"grad_norm": 7.613968849182129,
|
715 |
+
"learning_rate": 0.00023221848971300142,
|
716 |
+
"loss": 9.5994,
|
717 |
+
"step": 2000
|
718 |
+
},
|
719 |
+
{
|
720 |
+
"epoch": 1.1373873873873874,
|
721 |
+
"grad_norm": 8.049659729003906,
|
722 |
+
"learning_rate": 0.00023183041423211812,
|
723 |
+
"loss": 9.5675,
|
724 |
+
"step": 2020
|
725 |
+
},
|
726 |
+
{
|
727 |
+
"epoch": 1.1486486486486487,
|
728 |
+
"grad_norm": 8.708986282348633,
|
729 |
+
"learning_rate": 0.000231438481106822,
|
730 |
+
"loss": 9.2731,
|
731 |
+
"step": 2040
|
732 |
+
},
|
733 |
+
{
|
734 |
+
"epoch": 1.15990990990991,
|
735 |
+
"grad_norm": 8.820184707641602,
|
736 |
+
"learning_rate": 0.00023104270448981147,
|
737 |
+
"loss": 9.2091,
|
738 |
+
"step": 2060
|
739 |
+
},
|
740 |
+
{
|
741 |
+
"epoch": 1.1711711711711712,
|
742 |
+
"grad_norm": 8.059011459350586,
|
743 |
+
"learning_rate": 0.00023064309867257333,
|
744 |
+
"loss": 9.2435,
|
745 |
+
"step": 2080
|
746 |
+
},
|
747 |
+
{
|
748 |
+
"epoch": 1.1824324324324325,
|
749 |
+
"grad_norm": 8.836409568786621,
|
750 |
+
"learning_rate": 0.00023023967808486674,
|
751 |
+
"loss": 9.4296,
|
752 |
+
"step": 2100
|
753 |
+
},
|
754 |
+
{
|
755 |
+
"epoch": 1.1936936936936937,
|
756 |
+
"grad_norm": 7.625323295593262,
|
757 |
+
"learning_rate": 0.00022983245729420212,
|
758 |
+
"loss": 9.3096,
|
759 |
+
"step": 2120
|
760 |
+
},
|
761 |
+
{
|
762 |
+
"epoch": 1.204954954954955,
|
763 |
+
"grad_norm": 7.627326965332031,
|
764 |
+
"learning_rate": 0.00022942145100531526,
|
765 |
+
"loss": 9.1006,
|
766 |
+
"step": 2140
|
767 |
+
},
|
768 |
+
{
|
769 |
+
"epoch": 1.2162162162162162,
|
770 |
+
"grad_norm": 7.3621039390563965,
|
771 |
+
"learning_rate": 0.00022900667405963607,
|
772 |
+
"loss": 9.0271,
|
773 |
+
"step": 2160
|
774 |
+
},
|
775 |
+
{
|
776 |
+
"epoch": 1.2274774774774775,
|
777 |
+
"grad_norm": 10.079599380493164,
|
778 |
+
"learning_rate": 0.00022858814143475293,
|
779 |
+
"loss": 8.9905,
|
780 |
+
"step": 2180
|
781 |
+
},
|
782 |
+
{
|
783 |
+
"epoch": 1.2387387387387387,
|
784 |
+
"grad_norm": 7.299936771392822,
|
785 |
+
"learning_rate": 0.00022816586824387175,
|
786 |
+
"loss": 8.8307,
|
787 |
+
"step": 2200
|
788 |
+
},
|
789 |
+
{
|
790 |
+
"epoch": 1.25,
|
791 |
+
"grad_norm": 10.733438491821289,
|
792 |
+
"learning_rate": 0.00022773986973527005,
|
793 |
+
"loss": 8.7716,
|
794 |
+
"step": 2220
|
795 |
+
},
|
796 |
+
{
|
797 |
+
"epoch": 1.2612612612612613,
|
798 |
+
"grad_norm": 7.568991661071777,
|
799 |
+
"learning_rate": 0.0002273101612917467,
|
800 |
+
"loss": 8.9258,
|
801 |
+
"step": 2240
|
802 |
+
},
|
803 |
+
{
|
804 |
+
"epoch": 1.2725225225225225,
|
805 |
+
"grad_norm": 8.105926513671875,
|
806 |
+
"learning_rate": 0.00022687675843006616,
|
807 |
+
"loss": 8.5795,
|
808 |
+
"step": 2260
|
809 |
+
},
|
810 |
+
{
|
811 |
+
"epoch": 1.2837837837837838,
|
812 |
+
"grad_norm": 6.941417217254639,
|
813 |
+
"learning_rate": 0.0002264396768003983,
|
814 |
+
"loss": 8.7082,
|
815 |
+
"step": 2280
|
816 |
+
},
|
817 |
+
{
|
818 |
+
"epoch": 1.295045045045045,
|
819 |
+
"grad_norm": 7.045870780944824,
|
820 |
+
"learning_rate": 0.00022599893218575316,
|
821 |
+
"loss": 8.7914,
|
822 |
+
"step": 2300
|
823 |
+
},
|
824 |
+
{
|
825 |
+
"epoch": 1.3063063063063063,
|
826 |
+
"grad_norm": 8.087312698364258,
|
827 |
+
"learning_rate": 0.0002255545405014112,
|
828 |
+
"loss": 8.7433,
|
829 |
+
"step": 2320
|
830 |
+
},
|
831 |
+
{
|
832 |
+
"epoch": 1.3175675675675675,
|
833 |
+
"grad_norm": 8.753263473510742,
|
834 |
+
"learning_rate": 0.00022510651779434842,
|
835 |
+
"loss": 8.6449,
|
836 |
+
"step": 2340
|
837 |
+
},
|
838 |
+
{
|
839 |
+
"epoch": 1.3288288288288288,
|
840 |
+
"grad_norm": 8.395846366882324,
|
841 |
+
"learning_rate": 0.00022465488024265713,
|
842 |
+
"loss": 8.634,
|
843 |
+
"step": 2360
|
844 |
+
},
|
845 |
+
{
|
846 |
+
"epoch": 1.34009009009009,
|
847 |
+
"grad_norm": 7.994043350219727,
|
848 |
+
"learning_rate": 0.00022419964415496143,
|
849 |
+
"loss": 8.5735,
|
850 |
+
"step": 2380
|
851 |
+
},
|
852 |
+
{
|
853 |
+
"epoch": 1.3513513513513513,
|
854 |
+
"grad_norm": 8.119012832641602,
|
855 |
+
"learning_rate": 0.0002237408259698286,
|
856 |
+
"loss": 8.8715,
|
857 |
+
"step": 2400
|
858 |
+
},
|
859 |
+
{
|
860 |
+
"epoch": 1.3626126126126126,
|
861 |
+
"grad_norm": 7.6216583251953125,
|
862 |
+
"learning_rate": 0.00022327844225517547,
|
863 |
+
"loss": 8.3434,
|
864 |
+
"step": 2420
|
865 |
+
},
|
866 |
+
{
|
867 |
+
"epoch": 1.3738738738738738,
|
868 |
+
"grad_norm": 5.617091178894043,
|
869 |
+
"learning_rate": 0.00022281250970766984,
|
870 |
+
"loss": 8.5544,
|
871 |
+
"step": 2440
|
872 |
+
},
|
873 |
+
{
|
874 |
+
"epoch": 1.385135135135135,
|
875 |
+
"grad_norm": 7.126073360443115,
|
876 |
+
"learning_rate": 0.00022234304515212806,
|
877 |
+
"loss": 8.3322,
|
878 |
+
"step": 2460
|
879 |
+
},
|
880 |
+
{
|
881 |
+
"epoch": 1.3963963963963963,
|
882 |
+
"grad_norm": 7.028109073638916,
|
883 |
+
"learning_rate": 0.0002218700655409071,
|
884 |
+
"loss": 8.3188,
|
885 |
+
"step": 2480
|
886 |
+
},
|
887 |
+
{
|
888 |
+
"epoch": 1.4076576576576576,
|
889 |
+
"grad_norm": 7.848935127258301,
|
890 |
+
"learning_rate": 0.00022139358795329254,
|
891 |
+
"loss": 8.4047,
|
892 |
+
"step": 2500
|
893 |
+
},
|
894 |
+
{
|
895 |
+
"epoch": 1.4189189189189189,
|
896 |
+
"grad_norm": 7.840925693511963,
|
897 |
+
"learning_rate": 0.00022091362959488186,
|
898 |
+
"loss": 8.3978,
|
899 |
+
"step": 2520
|
900 |
+
},
|
901 |
+
{
|
902 |
+
"epoch": 1.4301801801801801,
|
903 |
+
"grad_norm": 7.0585198402404785,
|
904 |
+
"learning_rate": 0.00022043020779696304,
|
905 |
+
"loss": 8.2189,
|
906 |
+
"step": 2540
|
907 |
+
},
|
908 |
+
{
|
909 |
+
"epoch": 1.4414414414414414,
|
910 |
+
"grad_norm": 6.7776594161987305,
|
911 |
+
"learning_rate": 0.00021994334001588896,
|
912 |
+
"loss": 8.2129,
|
913 |
+
"step": 2560
|
914 |
+
},
|
915 |
+
{
|
916 |
+
"epoch": 1.4527027027027026,
|
917 |
+
"grad_norm": 6.949775218963623,
|
918 |
+
"learning_rate": 0.00021945304383244664,
|
919 |
+
"loss": 8.1717,
|
920 |
+
"step": 2580
|
921 |
+
},
|
922 |
+
{
|
923 |
+
"epoch": 1.4639639639639639,
|
924 |
+
"grad_norm": 6.041919708251953,
|
925 |
+
"learning_rate": 0.0002189593369512229,
|
926 |
+
"loss": 8.0273,
|
927 |
+
"step": 2600
|
928 |
+
},
|
929 |
+
{
|
930 |
+
"epoch": 1.4752252252252251,
|
931 |
+
"grad_norm": 7.787086486816406,
|
932 |
+
"learning_rate": 0.0002184622371999647,
|
933 |
+
"loss": 8.0657,
|
934 |
+
"step": 2620
|
935 |
+
},
|
936 |
+
{
|
937 |
+
"epoch": 1.4864864864864864,
|
938 |
+
"grad_norm": 6.464184761047363,
|
939 |
+
"learning_rate": 0.00021796176252893542,
|
940 |
+
"loss": 8.2728,
|
941 |
+
"step": 2640
|
942 |
+
},
|
943 |
+
{
|
944 |
+
"epoch": 1.4977477477477477,
|
945 |
+
"grad_norm": 6.4095354080200195,
|
946 |
+
"learning_rate": 0.0002174579310102668,
|
947 |
+
"loss": 7.9257,
|
948 |
+
"step": 2660
|
949 |
+
},
|
950 |
+
{
|
951 |
+
"epoch": 1.509009009009009,
|
952 |
+
"grad_norm": 7.329013824462891,
|
953 |
+
"learning_rate": 0.00021695076083730626,
|
954 |
+
"loss": 7.824,
|
955 |
+
"step": 2680
|
956 |
+
},
|
957 |
+
{
|
958 |
+
"epoch": 1.5202702702702702,
|
959 |
+
"grad_norm": 7.08155632019043,
|
960 |
+
"learning_rate": 0.00021644027032396003,
|
961 |
+
"loss": 7.9795,
|
962 |
+
"step": 2700
|
963 |
+
},
|
964 |
+
{
|
965 |
+
"epoch": 1.5315315315315314,
|
966 |
+
"grad_norm": 6.633808612823486,
|
967 |
+
"learning_rate": 0.00021592647790403172,
|
968 |
+
"loss": 7.9484,
|
969 |
+
"step": 2720
|
970 |
+
},
|
971 |
+
{
|
972 |
+
"epoch": 1.5427927927927927,
|
973 |
+
"grad_norm": 7.326924800872803,
|
974 |
+
"learning_rate": 0.0002154094021305567,
|
975 |
+
"loss": 8.0193,
|
976 |
+
"step": 2740
|
977 |
+
},
|
978 |
+
{
|
979 |
+
"epoch": 1.554054054054054,
|
980 |
+
"grad_norm": 6.548000812530518,
|
981 |
+
"learning_rate": 0.00021488906167513227,
|
982 |
+
"loss": 7.9684,
|
983 |
+
"step": 2760
|
984 |
+
},
|
985 |
+
{
|
986 |
+
"epoch": 1.5653153153153152,
|
987 |
+
"grad_norm": 6.106738090515137,
|
988 |
+
"learning_rate": 0.0002143654753272433,
|
989 |
+
"loss": 7.8051,
|
990 |
+
"step": 2780
|
991 |
+
},
|
992 |
+
{
|
993 |
+
"epoch": 1.5765765765765765,
|
994 |
+
"grad_norm": 6.181701183319092,
|
995 |
+
"learning_rate": 0.0002138386619935837,
|
996 |
+
"loss": 7.8747,
|
997 |
+
"step": 2800
|
998 |
+
},
|
999 |
+
{
|
1000 |
+
"epoch": 1.5878378378378377,
|
1001 |
+
"grad_norm": 6.695054054260254,
|
1002 |
+
"learning_rate": 0.00021330864069737387,
|
1003 |
+
"loss": 7.7717,
|
1004 |
+
"step": 2820
|
1005 |
+
},
|
1006 |
+
{
|
1007 |
+
"epoch": 1.599099099099099,
|
1008 |
+
"grad_norm": 7.034865379333496,
|
1009 |
+
"learning_rate": 0.0002127754305776737,
|
1010 |
+
"loss": 7.7304,
|
1011 |
+
"step": 2840
|
1012 |
+
},
|
1013 |
+
{
|
1014 |
+
"epoch": 1.6103603603603602,
|
1015 |
+
"grad_norm": 8.05134391784668,
|
1016 |
+
"learning_rate": 0.0002122390508886914,
|
1017 |
+
"loss": 7.8695,
|
1018 |
+
"step": 2860
|
1019 |
+
},
|
1020 |
+
{
|
1021 |
+
"epoch": 1.6216216216216215,
|
1022 |
+
"grad_norm": 6.66589879989624,
|
1023 |
+
"learning_rate": 0.0002116995209990882,
|
1024 |
+
"loss": 7.7153,
|
1025 |
+
"step": 2880
|
1026 |
+
},
|
1027 |
+
{
|
1028 |
+
"epoch": 1.6328828828828827,
|
1029 |
+
"grad_norm": 7.550106048583984,
|
1030 |
+
"learning_rate": 0.0002111568603912792,
|
1031 |
+
"loss": 7.6789,
|
1032 |
+
"step": 2900
|
1033 |
+
},
|
1034 |
+
{
|
1035 |
+
"epoch": 1.644144144144144,
|
1036 |
+
"grad_norm": 7.903695106506348,
|
1037 |
+
"learning_rate": 0.0002106110886607295,
|
1038 |
+
"loss": 7.6998,
|
1039 |
+
"step": 2920
|
1040 |
+
},
|
1041 |
+
{
|
1042 |
+
"epoch": 1.6554054054054053,
|
1043 |
+
"grad_norm": 7.171148300170898,
|
1044 |
+
"learning_rate": 0.0002100622255152469,
|
1045 |
+
"loss": 7.4489,
|
1046 |
+
"step": 2940
|
1047 |
+
},
|
1048 |
+
{
|
1049 |
+
"epoch": 1.6666666666666665,
|
1050 |
+
"grad_norm": 7.298549175262451,
|
1051 |
+
"learning_rate": 0.00020951029077427008,
|
1052 |
+
"loss": 7.558,
|
1053 |
+
"step": 2960
|
1054 |
+
},
|
1055 |
+
{
|
1056 |
+
"epoch": 1.6779279279279278,
|
1057 |
+
"grad_norm": 6.596391201019287,
|
1058 |
+
"learning_rate": 0.00020895530436815306,
|
1059 |
+
"loss": 7.6431,
|
1060 |
+
"step": 2980
|
1061 |
+
},
|
1062 |
+
{
|
1063 |
+
"epoch": 1.689189189189189,
|
1064 |
+
"grad_norm": 6.277204513549805,
|
1065 |
+
"learning_rate": 0.00020839728633744532,
|
1066 |
+
"loss": 7.6637,
|
1067 |
+
"step": 3000
|
1068 |
+
},
|
1069 |
+
{
|
1070 |
+
"epoch": 1.7004504504504503,
|
1071 |
+
"grad_norm": 5.395816802978516,
|
1072 |
+
"learning_rate": 0.00020783625683216828,
|
1073 |
+
"loss": 7.3164,
|
1074 |
+
"step": 3020
|
1075 |
+
},
|
1076 |
+
{
|
1077 |
+
"epoch": 1.7117117117117115,
|
1078 |
+
"grad_norm": 7.0829877853393555,
|
1079 |
+
"learning_rate": 0.00020727223611108778,
|
1080 |
+
"loss": 7.5608,
|
1081 |
+
"step": 3040
|
1082 |
+
},
|
1083 |
+
{
|
1084 |
+
"epoch": 1.722972972972973,
|
1085 |
+
"grad_norm": 6.532320976257324,
|
1086 |
+
"learning_rate": 0.00020670524454098228,
|
1087 |
+
"loss": 7.5514,
|
1088 |
+
"step": 3060
|
1089 |
+
},
|
1090 |
+
{
|
1091 |
+
"epoch": 1.7342342342342343,
|
1092 |
+
"grad_norm": 6.441926956176758,
|
1093 |
+
"learning_rate": 0.00020613530259590754,
|
1094 |
+
"loss": 7.4745,
|
1095 |
+
"step": 3080
|
1096 |
+
},
|
1097 |
+
{
|
1098 |
+
"epoch": 1.7454954954954955,
|
1099 |
+
"grad_norm": 6.796619415283203,
|
1100 |
+
"learning_rate": 0.00020556243085645743,
|
1101 |
+
"loss": 7.3947,
|
1102 |
+
"step": 3100
|
1103 |
+
},
|
1104 |
+
{
|
1105 |
+
"epoch": 1.7567567567567568,
|
1106 |
+
"grad_norm": 5.6059250831604,
|
1107 |
+
"learning_rate": 0.00020498665000902045,
|
1108 |
+
"loss": 7.3462,
|
1109 |
+
"step": 3120
|
1110 |
+
},
|
1111 |
+
{
|
1112 |
+
"epoch": 1.768018018018018,
|
1113 |
+
"grad_norm": 5.7582573890686035,
|
1114 |
+
"learning_rate": 0.00020440798084503321,
|
1115 |
+
"loss": 7.479,
|
1116 |
+
"step": 3140
|
1117 |
+
},
|
1118 |
+
{
|
1119 |
+
"epoch": 1.7792792792792793,
|
1120 |
+
"grad_norm": 6.546544551849365,
|
1121 |
+
"learning_rate": 0.0002038264442602291,
|
1122 |
+
"loss": 7.2541,
|
1123 |
+
"step": 3160
|
1124 |
+
},
|
1125 |
+
{
|
1126 |
+
"epoch": 1.7905405405405406,
|
1127 |
+
"grad_norm": 6.203800678253174,
|
1128 |
+
"learning_rate": 0.00020324206125388422,
|
1129 |
+
"loss": 7.313,
|
1130 |
+
"step": 3180
|
1131 |
+
},
|
1132 |
+
{
|
1133 |
+
"epoch": 1.8018018018018018,
|
1134 |
+
"grad_norm": 5.325755596160889,
|
1135 |
+
"learning_rate": 0.00020265485292805886,
|
1136 |
+
"loss": 7.2748,
|
1137 |
+
"step": 3200
|
1138 |
+
},
|
1139 |
+
{
|
1140 |
+
"epoch": 1.813063063063063,
|
1141 |
+
"grad_norm": 5.543078899383545,
|
1142 |
+
"learning_rate": 0.0002020648404868354,
|
1143 |
+
"loss": 7.3658,
|
1144 |
+
"step": 3220
|
1145 |
+
},
|
1146 |
+
{
|
1147 |
+
"epoch": 1.8243243243243243,
|
1148 |
+
"grad_norm": 6.311000347137451,
|
1149 |
+
"learning_rate": 0.00020147204523555303,
|
1150 |
+
"loss": 7.4673,
|
1151 |
+
"step": 3240
|
1152 |
+
},
|
1153 |
+
{
|
1154 |
+
"epoch": 1.8355855855855856,
|
1155 |
+
"grad_norm": 5.803033351898193,
|
1156 |
+
"learning_rate": 0.00020087648858003798,
|
1157 |
+
"loss": 7.3336,
|
1158 |
+
"step": 3260
|
1159 |
+
},
|
1160 |
+
{
|
1161 |
+
"epoch": 1.8468468468468469,
|
1162 |
+
"grad_norm": 5.618661403656006,
|
1163 |
+
"learning_rate": 0.0002002781920258308,
|
1164 |
+
"loss": 7.101,
|
1165 |
+
"step": 3280
|
1166 |
+
},
|
1167 |
+
{
|
1168 |
+
"epoch": 1.8581081081081081,
|
1169 |
+
"grad_norm": 5.574961185455322,
|
1170 |
+
"learning_rate": 0.00019967717717740975,
|
1171 |
+
"loss": 7.1956,
|
1172 |
+
"step": 3300
|
1173 |
+
},
|
1174 |
+
{
|
1175 |
+
"epoch": 1.8693693693693694,
|
1176 |
+
"grad_norm": 5.912872791290283,
|
1177 |
+
"learning_rate": 0.00019907346573741058,
|
1178 |
+
"loss": 7.0916,
|
1179 |
+
"step": 3320
|
1180 |
+
},
|
1181 |
+
{
|
1182 |
+
"epoch": 1.8806306306306306,
|
1183 |
+
"grad_norm": 9.942081451416016,
|
1184 |
+
"learning_rate": 0.00019846707950584299,
|
1185 |
+
"loss": 6.9485,
|
1186 |
+
"step": 3340
|
1187 |
+
},
|
1188 |
+
{
|
1189 |
+
"epoch": 1.8918918918918919,
|
1190 |
+
"grad_norm": 5.383841037750244,
|
1191 |
+
"learning_rate": 0.00019785804037930336,
|
1192 |
+
"loss": 7.1286,
|
1193 |
+
"step": 3360
|
1194 |
+
},
|
1195 |
+
{
|
1196 |
+
"epoch": 1.9031531531531531,
|
1197 |
+
"grad_norm": 6.271128177642822,
|
1198 |
+
"learning_rate": 0.00019724637035018394,
|
1199 |
+
"loss": 7.2622,
|
1200 |
+
"step": 3380
|
1201 |
+
},
|
1202 |
+
{
|
1203 |
+
"epoch": 1.9144144144144144,
|
1204 |
+
"grad_norm": 6.275426864624023,
|
1205 |
+
"learning_rate": 0.00019663209150587904,
|
1206 |
+
"loss": 7.179,
|
1207 |
+
"step": 3400
|
1208 |
+
},
|
1209 |
+
{
|
1210 |
+
"epoch": 1.9256756756756757,
|
1211 |
+
"grad_norm": 5.681616306304932,
|
1212 |
+
"learning_rate": 0.00019601522602798706,
|
1213 |
+
"loss": 6.9379,
|
1214 |
+
"step": 3420
|
1215 |
+
},
|
1216 |
+
{
|
1217 |
+
"epoch": 1.936936936936937,
|
1218 |
+
"grad_norm": 5.55466365814209,
|
1219 |
+
"learning_rate": 0.00019539579619150968,
|
1220 |
+
"loss": 6.9714,
|
1221 |
+
"step": 3440
|
1222 |
+
},
|
1223 |
+
{
|
1224 |
+
"epoch": 1.9481981981981982,
|
1225 |
+
"grad_norm": 5.896274566650391,
|
1226 |
+
"learning_rate": 0.0001947738243640477,
|
1227 |
+
"loss": 7.185,
|
1228 |
+
"step": 3460
|
1229 |
+
},
|
1230 |
+
{
|
1231 |
+
"epoch": 1.9594594594594594,
|
1232 |
+
"grad_norm": 6.699384689331055,
|
1233 |
+
"learning_rate": 0.00019414933300499303,
|
1234 |
+
"loss": 7.0395,
|
1235 |
+
"step": 3480
|
1236 |
+
},
|
1237 |
+
{
|
1238 |
+
"epoch": 1.9707207207207207,
|
1239 |
+
"grad_norm": 5.705533027648926,
|
1240 |
+
"learning_rate": 0.00019352234466471786,
|
1241 |
+
"loss": 7.0843,
|
1242 |
+
"step": 3500
|
1243 |
+
},
|
1244 |
+
{
|
1245 |
+
"epoch": 1.981981981981982,
|
1246 |
+
"grad_norm": 5.171501636505127,
|
1247 |
+
"learning_rate": 0.0001928928819837603,
|
1248 |
+
"loss": 7.3453,
|
1249 |
+
"step": 3520
|
1250 |
+
},
|
1251 |
+
{
|
1252 |
+
"epoch": 1.9932432432432432,
|
1253 |
+
"grad_norm": 5.449100971221924,
|
1254 |
+
"learning_rate": 0.00019226096769200696,
|
1255 |
+
"loss": 6.9518,
|
1256 |
+
"step": 3540
|
1257 |
+
},
|
1258 |
+
{
|
1259 |
+
"epoch": 2.0,
|
1260 |
+
"eval_loss": 6.912898063659668,
|
1261 |
+
"eval_runtime": 193.5223,
|
1262 |
+
"eval_samples_per_second": 42.331,
|
1263 |
+
"eval_steps_per_second": 5.291,
|
1264 |
+
"step": 3552
|
1265 |
+
},
|
1266 |
+
{
|
1267 |
+
"epoch": 2.0045045045045047,
|
1268 |
+
"grad_norm": 5.502616882324219,
|
1269 |
+
"learning_rate": 0.000191626624607872,
|
1270 |
+
"loss": 6.9544,
|
1271 |
+
"step": 3560
|
1272 |
+
},
|
1273 |
+
{
|
1274 |
+
"epoch": 2.015765765765766,
|
1275 |
+
"grad_norm": 7.196225166320801,
|
1276 |
+
"learning_rate": 0.0001909898756374731,
|
1277 |
+
"loss": 6.8782,
|
1278 |
+
"step": 3580
|
1279 |
+
},
|
1280 |
+
{
|
1281 |
+
"epoch": 2.027027027027027,
|
1282 |
+
"grad_norm": 5.475032806396484,
|
1283 |
+
"learning_rate": 0.0001903507437738047,
|
1284 |
+
"loss": 6.7123,
|
1285 |
+
"step": 3600
|
1286 |
+
},
|
1287 |
+
{
|
1288 |
+
"epoch": 2.0382882882882885,
|
1289 |
+
"grad_norm": 5.074850559234619,
|
1290 |
+
"learning_rate": 0.00018970925209590734,
|
1291 |
+
"loss": 6.7443,
|
1292 |
+
"step": 3620
|
1293 |
+
},
|
1294 |
+
{
|
1295 |
+
"epoch": 2.0495495495495497,
|
1296 |
+
"grad_norm": 5.7531328201293945,
|
1297 |
+
"learning_rate": 0.0001890654237680344,
|
1298 |
+
"loss": 6.7843,
|
1299 |
+
"step": 3640
|
1300 |
+
},
|
1301 |
+
{
|
1302 |
+
"epoch": 2.060810810810811,
|
1303 |
+
"grad_norm": 5.586028575897217,
|
1304 |
+
"learning_rate": 0.00018841928203881572,
|
1305 |
+
"loss": 6.6872,
|
1306 |
+
"step": 3660
|
1307 |
+
},
|
1308 |
+
{
|
1309 |
+
"epoch": 2.0720720720720722,
|
1310 |
+
"grad_norm": 5.3263936042785645,
|
1311 |
+
"learning_rate": 0.00018777085024041795,
|
1312 |
+
"loss": 6.7043,
|
1313 |
+
"step": 3680
|
1314 |
+
},
|
1315 |
+
{
|
1316 |
+
"epoch": 2.0833333333333335,
|
1317 |
+
"grad_norm": 6.154730319976807,
|
1318 |
+
"learning_rate": 0.00018712015178770218,
|
1319 |
+
"loss": 6.7238,
|
1320 |
+
"step": 3700
|
1321 |
+
},
|
1322 |
+
{
|
1323 |
+
"epoch": 2.0945945945945947,
|
1324 |
+
"grad_norm": 4.89910888671875,
|
1325 |
+
"learning_rate": 0.0001864672101773782,
|
1326 |
+
"loss": 6.6498,
|
1327 |
+
"step": 3720
|
1328 |
+
},
|
1329 |
+
{
|
1330 |
+
"epoch": 2.105855855855856,
|
1331 |
+
"grad_norm": 4.568950176239014,
|
1332 |
+
"learning_rate": 0.00018581204898715636,
|
1333 |
+
"loss": 6.689,
|
1334 |
+
"step": 3740
|
1335 |
+
},
|
1336 |
+
{
|
1337 |
+
"epoch": 2.1171171171171173,
|
1338 |
+
"grad_norm": 5.0181756019592285,
|
1339 |
+
"learning_rate": 0.00018515469187489595,
|
1340 |
+
"loss": 6.5897,
|
1341 |
+
"step": 3760
|
1342 |
+
},
|
1343 |
+
{
|
1344 |
+
"epoch": 2.1283783783783785,
|
1345 |
+
"grad_norm": 5.144893169403076,
|
1346 |
+
"learning_rate": 0.00018449516257775086,
|
1347 |
+
"loss": 6.6991,
|
1348 |
+
"step": 3780
|
1349 |
+
},
|
1350 |
+
{
|
1351 |
+
"epoch": 2.1396396396396398,
|
1352 |
+
"grad_norm": 6.296558856964111,
|
1353 |
+
"learning_rate": 0.00018383348491131268,
|
1354 |
+
"loss": 6.6091,
|
1355 |
+
"step": 3800
|
1356 |
+
},
|
1357 |
+
{
|
1358 |
+
"epoch": 2.150900900900901,
|
1359 |
+
"grad_norm": 5.029831886291504,
|
1360 |
+
"learning_rate": 0.00018316968276875048,
|
1361 |
+
"loss": 6.6498,
|
1362 |
+
"step": 3820
|
1363 |
+
},
|
1364 |
+
{
|
1365 |
+
"epoch": 2.1621621621621623,
|
1366 |
+
"grad_norm": 4.9656476974487305,
|
1367 |
+
"learning_rate": 0.00018250378011994817,
|
1368 |
+
"loss": 6.6533,
|
1369 |
+
"step": 3840
|
1370 |
+
},
|
1371 |
+
{
|
1372 |
+
"epoch": 2.1734234234234235,
|
1373 |
+
"grad_norm": 6.552328109741211,
|
1374 |
+
"learning_rate": 0.00018183580101063893,
|
1375 |
+
"loss": 6.6923,
|
1376 |
+
"step": 3860
|
1377 |
+
},
|
1378 |
+
{
|
1379 |
+
"epoch": 2.184684684684685,
|
1380 |
+
"grad_norm": 5.193049430847168,
|
1381 |
+
"learning_rate": 0.0001811657695615368,
|
1382 |
+
"loss": 6.5664,
|
1383 |
+
"step": 3880
|
1384 |
+
},
|
1385 |
+
{
|
1386 |
+
"epoch": 2.195945945945946,
|
1387 |
+
"grad_norm": 5.650617599487305,
|
1388 |
+
"learning_rate": 0.00018049370996746586,
|
1389 |
+
"loss": 6.7638,
|
1390 |
+
"step": 3900
|
1391 |
+
},
|
1392 |
+
{
|
1393 |
+
"epoch": 2.2072072072072073,
|
1394 |
+
"grad_norm": 4.744507789611816,
|
1395 |
+
"learning_rate": 0.0001798196464964865,
|
1396 |
+
"loss": 6.569,
|
1397 |
+
"step": 3920
|
1398 |
+
},
|
1399 |
+
{
|
1400 |
+
"epoch": 2.2184684684684686,
|
1401 |
+
"grad_norm": 4.649717330932617,
|
1402 |
+
"learning_rate": 0.00017914360348901892,
|
1403 |
+
"loss": 6.5865,
|
1404 |
+
"step": 3940
|
1405 |
+
},
|
1406 |
+
{
|
1407 |
+
"epoch": 2.22972972972973,
|
1408 |
+
"grad_norm": 5.418842315673828,
|
1409 |
+
"learning_rate": 0.00017846560535696455,
|
1410 |
+
"loss": 6.5317,
|
1411 |
+
"step": 3960
|
1412 |
+
},
|
1413 |
+
{
|
1414 |
+
"epoch": 2.240990990990991,
|
1415 |
+
"grad_norm": 5.146485805511475,
|
1416 |
+
"learning_rate": 0.00017778567658282407,
|
1417 |
+
"loss": 6.4661,
|
1418 |
+
"step": 3980
|
1419 |
+
},
|
1420 |
+
{
|
1421 |
+
"epoch": 2.2522522522522523,
|
1422 |
+
"grad_norm": 4.321829795837402,
|
1423 |
+
"learning_rate": 0.000177103841718814,
|
1424 |
+
"loss": 6.5814,
|
1425 |
+
"step": 4000
|
1426 |
+
},
|
1427 |
+
{
|
1428 |
+
"epoch": 2.2635135135135136,
|
1429 |
+
"grad_norm": 4.635056495666504,
|
1430 |
+
"learning_rate": 0.00017642012538597926,
|
1431 |
+
"loss": 6.5478,
|
1432 |
+
"step": 4020
|
1433 |
+
},
|
1434 |
+
{
|
1435 |
+
"epoch": 2.274774774774775,
|
1436 |
+
"grad_norm": 4.846772193908691,
|
1437 |
+
"learning_rate": 0.000175734552273305,
|
1438 |
+
"loss": 6.6611,
|
1439 |
+
"step": 4040
|
1440 |
+
},
|
1441 |
+
{
|
1442 |
+
"epoch": 2.286036036036036,
|
1443 |
+
"grad_norm": 4.857787609100342,
|
1444 |
+
"learning_rate": 0.00017504714713682438,
|
1445 |
+
"loss": 6.4596,
|
1446 |
+
"step": 4060
|
1447 |
+
},
|
1448 |
+
{
|
1449 |
+
"epoch": 2.2972972972972974,
|
1450 |
+
"grad_norm": 4.768703460693359,
|
1451 |
+
"learning_rate": 0.00017435793479872507,
|
1452 |
+
"loss": 6.508,
|
1453 |
+
"step": 4080
|
1454 |
+
},
|
1455 |
+
{
|
1456 |
+
"epoch": 2.3085585585585586,
|
1457 |
+
"grad_norm": 4.605207920074463,
|
1458 |
+
"learning_rate": 0.00017366694014645263,
|
1459 |
+
"loss": 6.3648,
|
1460 |
+
"step": 4100
|
1461 |
+
},
|
1462 |
+
{
|
1463 |
+
"epoch": 2.31981981981982,
|
1464 |
+
"grad_norm": 5.522574424743652,
|
1465 |
+
"learning_rate": 0.00017297418813181208,
|
1466 |
+
"loss": 6.5138,
|
1467 |
+
"step": 4120
|
1468 |
+
},
|
1469 |
+
{
|
1470 |
+
"epoch": 2.331081081081081,
|
1471 |
+
"grad_norm": 4.438038349151611,
|
1472 |
+
"learning_rate": 0.00017227970377006662,
|
1473 |
+
"loss": 6.4698,
|
1474 |
+
"step": 4140
|
1475 |
+
},
|
1476 |
+
{
|
1477 |
+
"epoch": 2.3423423423423424,
|
1478 |
+
"grad_norm": 5.221787929534912,
|
1479 |
+
"learning_rate": 0.00017158351213903466,
|
1480 |
+
"loss": 6.4,
|
1481 |
+
"step": 4160
|
1482 |
+
},
|
1483 |
+
{
|
1484 |
+
"epoch": 2.3536036036036037,
|
1485 |
+
"grad_norm": 5.3727498054504395,
|
1486 |
+
"learning_rate": 0.00017088563837818388,
|
1487 |
+
"loss": 6.472,
|
1488 |
+
"step": 4180
|
1489 |
+
},
|
1490 |
+
{
|
1491 |
+
"epoch": 2.364864864864865,
|
1492 |
+
"grad_norm": 4.393843650817871,
|
1493 |
+
"learning_rate": 0.00017018610768772385,
|
1494 |
+
"loss": 6.3323,
|
1495 |
+
"step": 4200
|
1496 |
+
},
|
1497 |
+
{
|
1498 |
+
"epoch": 2.376126126126126,
|
1499 |
+
"grad_norm": 4.410269260406494,
|
1500 |
+
"learning_rate": 0.0001694849453276956,
|
1501 |
+
"loss": 6.3458,
|
1502 |
+
"step": 4220
|
1503 |
+
},
|
1504 |
+
{
|
1505 |
+
"epoch": 2.3873873873873874,
|
1506 |
+
"grad_norm": 4.90555477142334,
|
1507 |
+
"learning_rate": 0.0001687821766170599,
|
1508 |
+
"loss": 6.3438,
|
1509 |
+
"step": 4240
|
1510 |
+
},
|
1511 |
+
{
|
1512 |
+
"epoch": 2.3986486486486487,
|
1513 |
+
"grad_norm": 4.615562915802002,
|
1514 |
+
"learning_rate": 0.00016807782693278273,
|
1515 |
+
"loss": 6.3853,
|
1516 |
+
"step": 4260
|
1517 |
+
},
|
1518 |
+
{
|
1519 |
+
"epoch": 2.40990990990991,
|
1520 |
+
"grad_norm": 4.589493274688721,
|
1521 |
+
"learning_rate": 0.00016737192170891902,
|
1522 |
+
"loss": 6.4002,
|
1523 |
+
"step": 4280
|
1524 |
+
},
|
1525 |
+
{
|
1526 |
+
"epoch": 2.421171171171171,
|
1527 |
+
"grad_norm": 4.222832679748535,
|
1528 |
+
"learning_rate": 0.00016666448643569426,
|
1529 |
+
"loss": 6.361,
|
1530 |
+
"step": 4300
|
1531 |
+
},
|
1532 |
+
{
|
1533 |
+
"epoch": 2.4324324324324325,
|
1534 |
+
"grad_norm": 5.57008171081543,
|
1535 |
+
"learning_rate": 0.00016595554665858387,
|
1536 |
+
"loss": 6.4508,
|
1537 |
+
"step": 4320
|
1538 |
+
},
|
1539 |
+
{
|
1540 |
+
"epoch": 2.4436936936936937,
|
1541 |
+
"grad_norm": 4.810327529907227,
|
1542 |
+
"learning_rate": 0.00016524512797739106,
|
1543 |
+
"loss": 6.3453,
|
1544 |
+
"step": 4340
|
1545 |
+
},
|
1546 |
+
{
|
1547 |
+
"epoch": 2.454954954954955,
|
1548 |
+
"grad_norm": 3.823190689086914,
|
1549 |
+
"learning_rate": 0.000164533256045322,
|
1550 |
+
"loss": 6.2972,
|
1551 |
+
"step": 4360
|
1552 |
+
},
|
1553 |
+
{
|
1554 |
+
"epoch": 2.4662162162162162,
|
1555 |
+
"grad_norm": 5.525811672210693,
|
1556 |
+
"learning_rate": 0.00016381995656806,
|
1557 |
+
"loss": 6.27,
|
1558 |
+
"step": 4380
|
1559 |
+
},
|
1560 |
+
{
|
1561 |
+
"epoch": 2.4774774774774775,
|
1562 |
+
"grad_norm": 4.199039459228516,
|
1563 |
+
"learning_rate": 0.00016310525530283682,
|
1564 |
+
"loss": 6.3205,
|
1565 |
+
"step": 4400
|
1566 |
+
},
|
1567 |
+
{
|
1568 |
+
"epoch": 2.4887387387387387,
|
1569 |
+
"grad_norm": 5.0115580558776855,
|
1570 |
+
"learning_rate": 0.00016238917805750275,
|
1571 |
+
"loss": 6.308,
|
1572 |
+
"step": 4420
|
1573 |
+
},
|
1574 |
+
{
|
1575 |
+
"epoch": 2.5,
|
1576 |
+
"grad_norm": 4.357995986938477,
|
1577 |
+
"learning_rate": 0.0001616717506895948,
|
1578 |
+
"loss": 6.2335,
|
1579 |
+
"step": 4440
|
1580 |
+
},
|
1581 |
+
{
|
1582 |
+
"epoch": 2.5112612612612613,
|
1583 |
+
"grad_norm": 5.8039374351501465,
|
1584 |
+
"learning_rate": 0.00016095299910540283,
|
1585 |
+
"loss": 6.1737,
|
1586 |
+
"step": 4460
|
1587 |
+
},
|
1588 |
+
{
|
1589 |
+
"epoch": 2.5225225225225225,
|
1590 |
+
"grad_norm": 4.064445972442627,
|
1591 |
+
"learning_rate": 0.00016023294925903412,
|
1592 |
+
"loss": 6.1518,
|
1593 |
+
"step": 4480
|
1594 |
+
},
|
1595 |
+
{
|
1596 |
+
"epoch": 2.5337837837837838,
|
1597 |
+
"grad_norm": 4.711318016052246,
|
1598 |
+
"learning_rate": 0.00015951162715147622,
|
1599 |
+
"loss": 6.2826,
|
1600 |
+
"step": 4500
|
1601 |
+
},
|
1602 |
+
{
|
1603 |
+
"epoch": 2.545045045045045,
|
1604 |
+
"grad_norm": 4.572793006896973,
|
1605 |
+
"learning_rate": 0.0001587890588296579,
|
1606 |
+
"loss": 6.1276,
|
1607 |
+
"step": 4520
|
1608 |
+
},
|
1609 |
+
{
|
1610 |
+
"epoch": 2.5563063063063063,
|
1611 |
+
"grad_norm": 5.005951881408691,
|
1612 |
+
"learning_rate": 0.00015806527038550884,
|
1613 |
+
"loss": 6.0659,
|
1614 |
+
"step": 4540
|
1615 |
+
},
|
1616 |
+
{
|
1617 |
+
"epoch": 2.5675675675675675,
|
1618 |
+
"grad_norm": 4.566344261169434,
|
1619 |
+
"learning_rate": 0.00015734028795501726,
|
1620 |
+
"loss": 6.2373,
|
1621 |
+
"step": 4560
|
1622 |
+
},
|
1623 |
+
{
|
1624 |
+
"epoch": 2.578828828828829,
|
1625 |
+
"grad_norm": 4.408982753753662,
|
1626 |
+
"learning_rate": 0.00015661413771728616,
|
1627 |
+
"loss": 6.1782,
|
1628 |
+
"step": 4580
|
1629 |
+
},
|
1630 |
+
{
|
1631 |
+
"epoch": 2.59009009009009,
|
1632 |
+
"grad_norm": 4.745791435241699,
|
1633 |
+
"learning_rate": 0.00015588684589358816,
|
1634 |
+
"loss": 6.1186,
|
1635 |
+
"step": 4600
|
1636 |
+
},
|
1637 |
+
{
|
1638 |
+
"epoch": 2.6013513513513513,
|
1639 |
+
"grad_norm": 4.748014450073242,
|
1640 |
+
"learning_rate": 0.00015515843874641834,
|
1641 |
+
"loss": 6.0719,
|
1642 |
+
"step": 4620
|
1643 |
+
},
|
1644 |
+
{
|
1645 |
+
"epoch": 2.6126126126126126,
|
1646 |
+
"grad_norm": 4.1786274909973145,
|
1647 |
+
"learning_rate": 0.00015442894257854632,
|
1648 |
+
"loss": 6.2196,
|
1649 |
+
"step": 4640
|
1650 |
+
},
|
1651 |
+
{
|
1652 |
+
"epoch": 2.623873873873874,
|
1653 |
+
"grad_norm": 5.255302906036377,
|
1654 |
+
"learning_rate": 0.00015369838373206602,
|
1655 |
+
"loss": 6.0788,
|
1656 |
+
"step": 4660
|
1657 |
+
},
|
1658 |
+
{
|
1659 |
+
"epoch": 2.635135135135135,
|
1660 |
+
"grad_norm": 5.268272399902344,
|
1661 |
+
"learning_rate": 0.00015296678858744487,
|
1662 |
+
"loss": 6.3061,
|
1663 |
+
"step": 4680
|
1664 |
+
},
|
1665 |
+
{
|
1666 |
+
"epoch": 2.6463963963963963,
|
1667 |
+
"grad_norm": 4.134327411651611,
|
1668 |
+
"learning_rate": 0.00015223418356257083,
|
1669 |
+
"loss": 6.1097,
|
1670 |
+
"step": 4700
|
1671 |
+
},
|
1672 |
+
{
|
1673 |
+
"epoch": 2.6576576576576576,
|
1674 |
+
"grad_norm": 5.097325325012207,
|
1675 |
+
"learning_rate": 0.00015150059511179873,
|
1676 |
+
"loss": 6.1005,
|
1677 |
+
"step": 4720
|
1678 |
+
},
|
1679 |
+
{
|
1680 |
+
"epoch": 2.668918918918919,
|
1681 |
+
"grad_norm": 5.016456604003906,
|
1682 |
+
"learning_rate": 0.00015076604972499483,
|
1683 |
+
"loss": 6.125,
|
1684 |
+
"step": 4740
|
1685 |
+
},
|
1686 |
+
{
|
1687 |
+
"epoch": 2.68018018018018,
|
1688 |
+
"grad_norm": 4.422632694244385,
|
1689 |
+
"learning_rate": 0.0001500305739265804,
|
1690 |
+
"loss": 6.0653,
|
1691 |
+
"step": 4760
|
1692 |
+
},
|
1693 |
+
{
|
1694 |
+
"epoch": 2.6914414414414414,
|
1695 |
+
"grad_norm": 4.201798915863037,
|
1696 |
+
"learning_rate": 0.00014929419427457373,
|
1697 |
+
"loss": 5.9961,
|
1698 |
+
"step": 4780
|
1699 |
+
},
|
1700 |
+
{
|
1701 |
+
"epoch": 2.7027027027027026,
|
1702 |
+
"grad_norm": 4.326559543609619,
|
1703 |
+
"learning_rate": 0.00014855693735963153,
|
1704 |
+
"loss": 6.104,
|
1705 |
+
"step": 4800
|
1706 |
+
},
|
1707 |
+
{
|
1708 |
+
"epoch": 2.713963963963964,
|
1709 |
+
"grad_norm": 8.840851783752441,
|
1710 |
+
"learning_rate": 0.00014781882980408805,
|
1711 |
+
"loss": 6.0508,
|
1712 |
+
"step": 4820
|
1713 |
+
},
|
1714 |
+
{
|
1715 |
+
"epoch": 2.725225225225225,
|
1716 |
+
"grad_norm": 4.185756683349609,
|
1717 |
+
"learning_rate": 0.0001470798982609945,
|
1718 |
+
"loss": 6.0414,
|
1719 |
+
"step": 4840
|
1720 |
+
},
|
1721 |
+
{
|
1722 |
+
"epoch": 2.7364864864864864,
|
1723 |
+
"grad_norm": 4.216666221618652,
|
1724 |
+
"learning_rate": 0.00014634016941315604,
|
1725 |
+
"loss": 5.8918,
|
1726 |
+
"step": 4860
|
1727 |
+
},
|
1728 |
+
{
|
1729 |
+
"epoch": 2.7477477477477477,
|
1730 |
+
"grad_norm": 4.4904608726501465,
|
1731 |
+
"learning_rate": 0.00014559966997216867,
|
1732 |
+
"loss": 6.0668,
|
1733 |
+
"step": 4880
|
1734 |
+
},
|
1735 |
+
{
|
1736 |
+
"epoch": 2.759009009009009,
|
1737 |
+
"grad_norm": 4.338995933532715,
|
1738 |
+
"learning_rate": 0.0001448584266774544,
|
1739 |
+
"loss": 5.9947,
|
1740 |
+
"step": 4900
|
1741 |
+
},
|
1742 |
+
{
|
1743 |
+
"epoch": 2.77027027027027,
|
1744 |
+
"grad_norm": 6.664434432983398,
|
1745 |
+
"learning_rate": 0.0001441164662952958,
|
1746 |
+
"loss": 5.9901,
|
1747 |
+
"step": 4920
|
1748 |
+
},
|
1749 |
+
{
|
1750 |
+
"epoch": 2.7815315315315314,
|
1751 |
+
"grad_norm": 3.878373622894287,
|
1752 |
+
"learning_rate": 0.00014337381561786944,
|
1753 |
+
"loss": 6.0326,
|
1754 |
+
"step": 4940
|
1755 |
+
},
|
1756 |
+
{
|
1757 |
+
"epoch": 2.7927927927927927,
|
1758 |
+
"grad_norm": 4.18255615234375,
|
1759 |
+
"learning_rate": 0.0001426305014622784,
|
1760 |
+
"loss": 6.0268,
|
1761 |
+
"step": 4960
|
1762 |
+
},
|
1763 |
+
{
|
1764 |
+
"epoch": 2.804054054054054,
|
1765 |
+
"grad_norm": 4.199193000793457,
|
1766 |
+
"learning_rate": 0.00014188655066958415,
|
1767 |
+
"loss": 5.9767,
|
1768 |
+
"step": 4980
|
1769 |
+
},
|
1770 |
+
{
|
1771 |
+
"epoch": 2.815315315315315,
|
1772 |
+
"grad_norm": 4.787850379943848,
|
1773 |
+
"learning_rate": 0.00014114199010383682,
|
1774 |
+
"loss": 6.0231,
|
1775 |
+
"step": 5000
|
1776 |
+
},
|
1777 |
+
{
|
1778 |
+
"epoch": 2.8265765765765765,
|
1779 |
+
"grad_norm": 5.528329849243164,
|
1780 |
+
"learning_rate": 0.00014039684665110568,
|
1781 |
+
"loss": 5.9916,
|
1782 |
+
"step": 5020
|
1783 |
+
},
|
1784 |
+
{
|
1785 |
+
"epoch": 2.8378378378378377,
|
1786 |
+
"grad_norm": 4.42682409286499,
|
1787 |
+
"learning_rate": 0.0001396511472185079,
|
1788 |
+
"loss": 5.9346,
|
1789 |
+
"step": 5040
|
1790 |
+
},
|
1791 |
+
{
|
1792 |
+
"epoch": 2.849099099099099,
|
1793 |
+
"grad_norm": 4.8591628074646,
|
1794 |
+
"learning_rate": 0.00013890491873323708,
|
1795 |
+
"loss": 6.0186,
|
1796 |
+
"step": 5060
|
1797 |
+
},
|
1798 |
+
{
|
1799 |
+
"epoch": 2.8603603603603602,
|
1800 |
+
"grad_norm": 4.678809642791748,
|
1801 |
+
"learning_rate": 0.00013815818814159092,
|
1802 |
+
"loss": 5.9332,
|
1803 |
+
"step": 5080
|
1804 |
+
},
|
1805 |
+
{
|
1806 |
+
"epoch": 2.8716216216216215,
|
1807 |
+
"grad_norm": 4.1150336265563965,
|
1808 |
+
"learning_rate": 0.00013741098240799832,
|
1809 |
+
"loss": 6.1274,
|
1810 |
+
"step": 5100
|
1811 |
+
},
|
1812 |
+
{
|
1813 |
+
"epoch": 2.8828828828828827,
|
1814 |
+
"grad_norm": 4.497319221496582,
|
1815 |
+
"learning_rate": 0.00013666332851404526,
|
1816 |
+
"loss": 5.9654,
|
1817 |
+
"step": 5120
|
1818 |
+
},
|
1819 |
+
{
|
1820 |
+
"epoch": 2.894144144144144,
|
1821 |
+
"grad_norm": 4.345571041107178,
|
1822 |
+
"learning_rate": 0.00013591525345750102,
|
1823 |
+
"loss": 5.9232,
|
1824 |
+
"step": 5140
|
1825 |
+
},
|
1826 |
+
{
|
1827 |
+
"epoch": 2.9054054054054053,
|
1828 |
+
"grad_norm": 4.747057914733887,
|
1829 |
+
"learning_rate": 0.00013516678425134294,
|
1830 |
+
"loss": 6.0654,
|
1831 |
+
"step": 5160
|
1832 |
+
},
|
1833 |
+
{
|
1834 |
+
"epoch": 2.9166666666666665,
|
1835 |
+
"grad_norm": 4.389774322509766,
|
1836 |
+
"learning_rate": 0.00013441794792278113,
|
1837 |
+
"loss": 5.879,
|
1838 |
+
"step": 5180
|
1839 |
+
},
|
1840 |
+
{
|
1841 |
+
"epoch": 2.9279279279279278,
|
1842 |
+
"grad_norm": 3.92946195602417,
|
1843 |
+
"learning_rate": 0.00013366877151228254,
|
1844 |
+
"loss": 5.9097,
|
1845 |
+
"step": 5200
|
1846 |
+
},
|
1847 |
+
{
|
1848 |
+
"epoch": 2.939189189189189,
|
1849 |
+
"grad_norm": 4.358794689178467,
|
1850 |
+
"learning_rate": 0.00013291928207259432,
|
1851 |
+
"loss": 5.9055,
|
1852 |
+
"step": 5220
|
1853 |
+
},
|
1854 |
+
{
|
1855 |
+
"epoch": 2.9504504504504503,
|
1856 |
+
"grad_norm": 4.1789984703063965,
|
1857 |
+
"learning_rate": 0.00013216950666776725,
|
1858 |
+
"loss": 5.7307,
|
1859 |
+
"step": 5240
|
1860 |
+
},
|
1861 |
+
{
|
1862 |
+
"epoch": 2.9617117117117115,
|
1863 |
+
"grad_norm": 4.347870826721191,
|
1864 |
+
"learning_rate": 0.0001314194723721782,
|
1865 |
+
"loss": 5.8359,
|
1866 |
+
"step": 5260
|
1867 |
+
},
|
1868 |
+
{
|
1869 |
+
"epoch": 2.972972972972973,
|
1870 |
+
"grad_norm": 4.1626362800598145,
|
1871 |
+
"learning_rate": 0.00013066920626955277,
|
1872 |
+
"loss": 5.7717,
|
1873 |
+
"step": 5280
|
1874 |
+
},
|
1875 |
+
{
|
1876 |
+
"epoch": 2.9842342342342345,
|
1877 |
+
"grad_norm": 4.5532145500183105,
|
1878 |
+
"learning_rate": 0.00012991873545198674,
|
1879 |
+
"loss": 5.8688,
|
1880 |
+
"step": 5300
|
1881 |
+
},
|
1882 |
+
{
|
1883 |
+
"epoch": 2.9954954954954953,
|
1884 |
+
"grad_norm": 4.3807525634765625,
|
1885 |
+
"learning_rate": 0.00012916808701896854,
|
1886 |
+
"loss": 5.7895,
|
1887 |
+
"step": 5320
|
1888 |
+
},
|
1889 |
+
{
|
1890 |
+
"epoch": 3.0,
|
1891 |
+
"eval_loss": 5.978276252746582,
|
1892 |
+
"eval_runtime": 193.4861,
|
1893 |
+
"eval_samples_per_second": 42.339,
|
1894 |
+
"eval_steps_per_second": 5.292,
|
1895 |
+
"step": 5328
|
1896 |
+
},
|
1897 |
+
{
|
1898 |
+
"epoch": 3.0067567567567566,
|
1899 |
+
"grad_norm": 4.323426723480225,
|
1900 |
+
"learning_rate": 0.00012841728807640008,
|
1901 |
+
"loss": 5.7336,
|
1902 |
+
"step": 5340
|
1903 |
+
},
|
1904 |
+
{
|
1905 |
+
"epoch": 3.018018018018018,
|
1906 |
+
"grad_norm": 4.7407660484313965,
|
1907 |
+
"learning_rate": 0.00012766636573561822,
|
1908 |
+
"loss": 5.6844,
|
1909 |
+
"step": 5360
|
1910 |
+
},
|
1911 |
+
{
|
1912 |
+
"epoch": 3.029279279279279,
|
1913 |
+
"grad_norm": 4.333187103271484,
|
1914 |
+
"learning_rate": 0.00012691534711241567,
|
1915 |
+
"loss": 5.8184,
|
1916 |
+
"step": 5380
|
1917 |
+
},
|
1918 |
+
{
|
1919 |
+
"epoch": 3.0405405405405403,
|
1920 |
+
"grad_norm": 3.732142925262451,
|
1921 |
+
"learning_rate": 0.000126164259326062,
|
1922 |
+
"loss": 5.7597,
|
1923 |
+
"step": 5400
|
1924 |
+
},
|
1925 |
+
{
|
1926 |
+
"epoch": 3.0518018018018016,
|
1927 |
+
"grad_norm": 3.862643241882324,
|
1928 |
+
"learning_rate": 0.00012541312949832412,
|
1929 |
+
"loss": 5.659,
|
1930 |
+
"step": 5420
|
1931 |
+
},
|
1932 |
+
{
|
1933 |
+
"epoch": 3.063063063063063,
|
1934 |
+
"grad_norm": 4.887398719787598,
|
1935 |
+
"learning_rate": 0.00012466198475248721,
|
1936 |
+
"loss": 5.8555,
|
1937 |
+
"step": 5440
|
1938 |
+
},
|
1939 |
+
{
|
1940 |
+
"epoch": 3.074324324324324,
|
1941 |
+
"grad_norm": 3.9390506744384766,
|
1942 |
+
"learning_rate": 0.00012391085221237508,
|
1943 |
+
"loss": 5.736,
|
1944 |
+
"step": 5460
|
1945 |
+
},
|
1946 |
+
{
|
1947 |
+
"epoch": 3.0855855855855854,
|
1948 |
+
"grad_norm": 4.177529335021973,
|
1949 |
+
"learning_rate": 0.00012315975900137072,
|
1950 |
+
"loss": 5.7396,
|
1951 |
+
"step": 5480
|
1952 |
+
},
|
1953 |
+
{
|
1954 |
+
"epoch": 3.0968468468468466,
|
1955 |
+
"grad_norm": 4.195279121398926,
|
1956 |
+
"learning_rate": 0.00012240873224143704,
|
1957 |
+
"loss": 5.6204,
|
1958 |
+
"step": 5500
|
1959 |
+
},
|
1960 |
+
{
|
1961 |
+
"epoch": 3.108108108108108,
|
1962 |
+
"grad_norm": 4.1104631423950195,
|
1963 |
+
"learning_rate": 0.00012165779905213741,
|
1964 |
+
"loss": 5.7094,
|
1965 |
+
"step": 5520
|
1966 |
+
},
|
1967 |
+
{
|
1968 |
+
"epoch": 3.1193693693693696,
|
1969 |
+
"grad_norm": 3.631848096847534,
|
1970 |
+
"learning_rate": 0.00012090698654965624,
|
1971 |
+
"loss": 5.6496,
|
1972 |
+
"step": 5540
|
1973 |
+
},
|
1974 |
+
{
|
1975 |
+
"epoch": 3.1306306306306304,
|
1976 |
+
"grad_norm": 3.695162057876587,
|
1977 |
+
"learning_rate": 0.00012015632184582009,
|
1978 |
+
"loss": 5.6944,
|
1979 |
+
"step": 5560
|
1980 |
+
},
|
1981 |
+
{
|
1982 |
+
"epoch": 3.141891891891892,
|
1983 |
+
"grad_norm": 3.7635786533355713,
|
1984 |
+
"learning_rate": 0.00011940583204711839,
|
1985 |
+
"loss": 5.6392,
|
1986 |
+
"step": 5580
|
1987 |
+
},
|
1988 |
+
{
|
1989 |
+
"epoch": 3.153153153153153,
|
1990 |
+
"grad_norm": 3.6323797702789307,
|
1991 |
+
"learning_rate": 0.00011865554425372489,
|
1992 |
+
"loss": 5.6936,
|
1993 |
+
"step": 5600
|
1994 |
+
},
|
1995 |
+
{
|
1996 |
+
"epoch": 3.1644144144144146,
|
1997 |
+
"grad_norm": 4.534457206726074,
|
1998 |
+
"learning_rate": 0.00011790548555851882,
|
1999 |
+
"loss": 5.5895,
|
2000 |
+
"step": 5620
|
2001 |
+
},
|
2002 |
+
{
|
2003 |
+
"epoch": 3.175675675675676,
|
2004 |
+
"grad_norm": 3.876401662826538,
|
2005 |
+
"learning_rate": 0.00011715568304610665,
|
2006 |
+
"loss": 5.5675,
|
2007 |
+
"step": 5640
|
2008 |
+
},
|
2009 |
+
{
|
2010 |
+
"epoch": 3.186936936936937,
|
2011 |
+
"grad_norm": 4.627795696258545,
|
2012 |
+
"learning_rate": 0.00011640616379184423,
|
2013 |
+
"loss": 5.7316,
|
2014 |
+
"step": 5660
|
2015 |
+
},
|
2016 |
+
{
|
2017 |
+
"epoch": 3.1981981981981984,
|
2018 |
+
"grad_norm": 4.265388011932373,
|
2019 |
+
"learning_rate": 0.00011565695486085888,
|
2020 |
+
"loss": 5.6629,
|
2021 |
+
"step": 5680
|
2022 |
+
},
|
2023 |
+
{
|
2024 |
+
"epoch": 3.2094594594594597,
|
2025 |
+
"grad_norm": 4.490591049194336,
|
2026 |
+
"learning_rate": 0.00011490808330707209,
|
2027 |
+
"loss": 5.5652,
|
2028 |
+
"step": 5700
|
2029 |
+
},
|
2030 |
+
{
|
2031 |
+
"epoch": 3.220720720720721,
|
2032 |
+
"grad_norm": 4.579701900482178,
|
2033 |
+
"learning_rate": 0.00011415957617222281,
|
2034 |
+
"loss": 5.4817,
|
2035 |
+
"step": 5720
|
2036 |
+
},
|
2037 |
+
{
|
2038 |
+
"epoch": 3.231981981981982,
|
2039 |
+
"grad_norm": 4.159543991088867,
|
2040 |
+
"learning_rate": 0.00011341146048489073,
|
2041 |
+
"loss": 5.613,
|
2042 |
+
"step": 5740
|
2043 |
+
},
|
2044 |
+
{
|
2045 |
+
"epoch": 3.2432432432432434,
|
2046 |
+
"grad_norm": 3.850271701812744,
|
2047 |
+
"learning_rate": 0.00011266376325952038,
|
2048 |
+
"loss": 5.6695,
|
2049 |
+
"step": 5760
|
2050 |
+
},
|
2051 |
+
{
|
2052 |
+
"epoch": 3.2545045045045047,
|
2053 |
+
"grad_norm": 3.8045921325683594,
|
2054 |
+
"learning_rate": 0.00011191651149544564,
|
2055 |
+
"loss": 5.6224,
|
2056 |
+
"step": 5780
|
2057 |
+
},
|
2058 |
+
{
|
2059 |
+
"epoch": 3.265765765765766,
|
2060 |
+
"grad_norm": 3.819220781326294,
|
2061 |
+
"learning_rate": 0.00011116973217591485,
|
2062 |
+
"loss": 5.6162,
|
2063 |
+
"step": 5800
|
2064 |
+
},
|
2065 |
+
{
|
2066 |
+
"epoch": 3.277027027027027,
|
2067 |
+
"grad_norm": 3.965785503387451,
|
2068 |
+
"learning_rate": 0.00011042345226711632,
|
2069 |
+
"loss": 5.6112,
|
2070 |
+
"step": 5820
|
2071 |
+
},
|
2072 |
+
{
|
2073 |
+
"epoch": 3.2882882882882885,
|
2074 |
+
"grad_norm": 4.823921203613281,
|
2075 |
+
"learning_rate": 0.00010967769871720463,
|
2076 |
+
"loss": 5.5894,
|
2077 |
+
"step": 5840
|
2078 |
+
},
|
2079 |
+
{
|
2080 |
+
"epoch": 3.2995495495495497,
|
2081 |
+
"grad_norm": 3.6114063262939453,
|
2082 |
+
"learning_rate": 0.00010893249845532768,
|
2083 |
+
"loss": 5.578,
|
2084 |
+
"step": 5860
|
2085 |
+
},
|
2086 |
+
{
|
2087 |
+
"epoch": 3.310810810810811,
|
2088 |
+
"grad_norm": 3.837531089782715,
|
2089 |
+
"learning_rate": 0.00010818787839065398,
|
2090 |
+
"loss": 5.5494,
|
2091 |
+
"step": 5880
|
2092 |
+
},
|
2093 |
+
{
|
2094 |
+
"epoch": 3.3220720720720722,
|
2095 |
+
"grad_norm": 4.008458137512207,
|
2096 |
+
"learning_rate": 0.00010744386541140123,
|
2097 |
+
"loss": 5.544,
|
2098 |
+
"step": 5900
|
2099 |
+
},
|
2100 |
+
{
|
2101 |
+
"epoch": 3.3333333333333335,
|
2102 |
+
"grad_norm": 3.6150972843170166,
|
2103 |
+
"learning_rate": 0.00010670048638386522,
|
2104 |
+
"loss": 5.641,
|
2105 |
+
"step": 5920
|
2106 |
+
},
|
2107 |
+
{
|
2108 |
+
"epoch": 3.3445945945945947,
|
2109 |
+
"grad_norm": 3.761298418045044,
|
2110 |
+
"learning_rate": 0.00010595776815144985,
|
2111 |
+
"loss": 5.5959,
|
2112 |
+
"step": 5940
|
2113 |
+
},
|
2114 |
+
{
|
2115 |
+
"epoch": 3.355855855855856,
|
2116 |
+
"grad_norm": 3.985516309738159,
|
2117 |
+
"learning_rate": 0.00010521573753369768,
|
2118 |
+
"loss": 5.5627,
|
2119 |
+
"step": 5960
|
2120 |
+
},
|
2121 |
+
{
|
2122 |
+
"epoch": 3.3671171171171173,
|
2123 |
+
"grad_norm": 3.49960994720459,
|
2124 |
+
"learning_rate": 0.00010447442132532137,
|
2125 |
+
"loss": 5.5396,
|
2126 |
+
"step": 5980
|
2127 |
+
},
|
2128 |
+
{
|
2129 |
+
"epoch": 3.3783783783783785,
|
2130 |
+
"grad_norm": 4.1534104347229,
|
2131 |
+
"learning_rate": 0.00010373384629523657,
|
2132 |
+
"loss": 5.3613,
|
2133 |
+
"step": 6000
|
2134 |
+
},
|
2135 |
+
{
|
2136 |
+
"epoch": 3.3896396396396398,
|
2137 |
+
"grad_norm": 3.6914377212524414,
|
2138 |
+
"learning_rate": 0.00010299403918559476,
|
2139 |
+
"loss": 5.4549,
|
2140 |
+
"step": 6020
|
2141 |
+
},
|
2142 |
+
{
|
2143 |
+
"epoch": 3.400900900900901,
|
2144 |
+
"grad_norm": 3.396759510040283,
|
2145 |
+
"learning_rate": 0.00010225502671081782,
|
2146 |
+
"loss": 5.4082,
|
2147 |
+
"step": 6040
|
2148 |
+
},
|
2149 |
+
{
|
2150 |
+
"epoch": 3.4121621621621623,
|
2151 |
+
"grad_norm": 3.7566490173339844,
|
2152 |
+
"learning_rate": 0.00010151683555663356,
|
2153 |
+
"loss": 5.5457,
|
2154 |
+
"step": 6060
|
2155 |
+
},
|
2156 |
+
{
|
2157 |
+
"epoch": 3.4234234234234235,
|
2158 |
+
"grad_norm": 4.450892448425293,
|
2159 |
+
"learning_rate": 0.0001007794923791118,
|
2160 |
+
"loss": 5.5,
|
2161 |
+
"step": 6080
|
2162 |
+
},
|
2163 |
+
{
|
2164 |
+
"epoch": 3.434684684684685,
|
2165 |
+
"grad_norm": 3.8145172595977783,
|
2166 |
+
"learning_rate": 0.00010004302380370185,
|
2167 |
+
"loss": 5.4815,
|
2168 |
+
"step": 6100
|
2169 |
+
},
|
2170 |
+
{
|
2171 |
+
"epoch": 3.445945945945946,
|
2172 |
+
"grad_norm": 3.273996114730835,
|
2173 |
+
"learning_rate": 9.930745642427129e-05,
|
2174 |
+
"loss": 5.5172,
|
2175 |
+
"step": 6120
|
2176 |
+
},
|
2177 |
+
{
|
2178 |
+
"epoch": 3.4572072072072073,
|
2179 |
+
"grad_norm": 4.224421977996826,
|
2180 |
+
"learning_rate": 9.857281680214552e-05,
|
2181 |
+
"loss": 5.4802,
|
2182 |
+
"step": 6140
|
2183 |
+
},
|
2184 |
+
{
|
2185 |
+
"epoch": 3.4684684684684686,
|
2186 |
+
"grad_norm": 3.935595989227295,
|
2187 |
+
"learning_rate": 9.783913146514862e-05,
|
2188 |
+
"loss": 5.4382,
|
2189 |
+
"step": 6160
|
2190 |
+
},
|
2191 |
+
{
|
2192 |
+
"epoch": 3.47972972972973,
|
2193 |
+
"grad_norm": 3.488819122314453,
|
2194 |
+
"learning_rate": 9.710642690664534e-05,
|
2195 |
+
"loss": 5.4057,
|
2196 |
+
"step": 6180
|
2197 |
+
},
|
2198 |
+
{
|
2199 |
+
"epoch": 3.490990990990991,
|
2200 |
+
"grad_norm": 3.711963176727295,
|
2201 |
+
"learning_rate": 9.637472958458475e-05,
|
2202 |
+
"loss": 5.3696,
|
2203 |
+
"step": 6200
|
2204 |
+
},
|
2205 |
+
{
|
2206 |
+
"epoch": 3.5022522522522523,
|
2207 |
+
"grad_norm": 4.166729927062988,
|
2208 |
+
"learning_rate": 9.564406592054448e-05,
|
2209 |
+
"loss": 5.4406,
|
2210 |
+
"step": 6220
|
2211 |
+
},
|
2212 |
+
{
|
2213 |
+
"epoch": 3.5135135135135136,
|
2214 |
+
"grad_norm": 3.605489492416382,
|
2215 |
+
"learning_rate": 9.491446229877674e-05,
|
2216 |
+
"loss": 5.4604,
|
2217 |
+
"step": 6240
|
2218 |
+
},
|
2219 |
+
{
|
2220 |
+
"epoch": 3.524774774774775,
|
2221 |
+
"grad_norm": 3.7834997177124023,
|
2222 |
+
"learning_rate": 9.41859450652558e-05,
|
2223 |
+
"loss": 5.4426,
|
2224 |
+
"step": 6260
|
2225 |
+
},
|
2226 |
+
{
|
2227 |
+
"epoch": 3.536036036036036,
|
2228 |
+
"grad_norm": 3.863955020904541,
|
2229 |
+
"learning_rate": 9.345854052672635e-05,
|
2230 |
+
"loss": 5.3948,
|
2231 |
+
"step": 6280
|
2232 |
+
},
|
2233 |
+
{
|
2234 |
+
"epoch": 3.5472972972972974,
|
2235 |
+
"grad_norm": 3.517244577407837,
|
2236 |
+
"learning_rate": 9.273227494975375e-05,
|
2237 |
+
"loss": 5.3699,
|
2238 |
+
"step": 6300
|
2239 |
+
},
|
2240 |
+
{
|
2241 |
+
"epoch": 3.5585585585585586,
|
2242 |
+
"grad_norm": 3.529064416885376,
|
2243 |
+
"learning_rate": 9.200717455977531e-05,
|
2244 |
+
"loss": 5.2605,
|
2245 |
+
"step": 6320
|
2246 |
+
},
|
2247 |
+
{
|
2248 |
+
"epoch": 3.56981981981982,
|
2249 |
+
"grad_norm": 3.515017509460449,
|
2250 |
+
"learning_rate": 9.128326554015376e-05,
|
2251 |
+
"loss": 5.4202,
|
2252 |
+
"step": 6340
|
2253 |
+
},
|
2254 |
+
{
|
2255 |
+
"epoch": 3.581081081081081,
|
2256 |
+
"grad_norm": 3.1703290939331055,
|
2257 |
+
"learning_rate": 9.056057403123119e-05,
|
2258 |
+
"loss": 5.2899,
|
2259 |
+
"step": 6360
|
2260 |
+
},
|
2261 |
+
{
|
2262 |
+
"epoch": 3.5923423423423424,
|
2263 |
+
"grad_norm": 3.3577380180358887,
|
2264 |
+
"learning_rate": 8.983912612938552e-05,
|
2265 |
+
"loss": 5.3383,
|
2266 |
+
"step": 6380
|
2267 |
+
},
|
2268 |
+
{
|
2269 |
+
"epoch": 3.6036036036036037,
|
2270 |
+
"grad_norm": 3.441986560821533,
|
2271 |
+
"learning_rate": 8.911894788608806e-05,
|
2272 |
+
"loss": 5.3425,
|
2273 |
+
"step": 6400
|
2274 |
+
},
|
2275 |
+
{
|
2276 |
+
"epoch": 3.614864864864865,
|
2277 |
+
"grad_norm": 3.591794967651367,
|
2278 |
+
"learning_rate": 8.840006530696274e-05,
|
2279 |
+
"loss": 5.4011,
|
2280 |
+
"step": 6420
|
2281 |
+
},
|
2282 |
+
{
|
2283 |
+
"epoch": 3.626126126126126,
|
2284 |
+
"grad_norm": 3.424859046936035,
|
2285 |
+
"learning_rate": 8.768250435084695e-05,
|
2286 |
+
"loss": 5.2703,
|
2287 |
+
"step": 6440
|
2288 |
+
},
|
2289 |
+
{
|
2290 |
+
"epoch": 3.6373873873873874,
|
2291 |
+
"grad_norm": 3.5369760990142822,
|
2292 |
+
"learning_rate": 8.696629092885464e-05,
|
2293 |
+
"loss": 5.2353,
|
2294 |
+
"step": 6460
|
2295 |
+
},
|
2296 |
+
{
|
2297 |
+
"epoch": 3.6486486486486487,
|
2298 |
+
"grad_norm": 3.3173818588256836,
|
2299 |
+
"learning_rate": 8.625145090343994e-05,
|
2300 |
+
"loss": 5.3336,
|
2301 |
+
"step": 6480
|
2302 |
+
},
|
2303 |
+
{
|
2304 |
+
"epoch": 3.65990990990991,
|
2305 |
+
"grad_norm": 3.555009126663208,
|
2306 |
+
"learning_rate": 8.553801008746388e-05,
|
2307 |
+
"loss": 5.3216,
|
2308 |
+
"step": 6500
|
2309 |
+
},
|
2310 |
+
{
|
2311 |
+
"epoch": 3.671171171171171,
|
2312 |
+
"grad_norm": 3.4357593059539795,
|
2313 |
+
"learning_rate": 8.482599424326197e-05,
|
2314 |
+
"loss": 5.2752,
|
2315 |
+
"step": 6520
|
2316 |
+
},
|
2317 |
+
{
|
2318 |
+
"epoch": 3.6824324324324325,
|
2319 |
+
"grad_norm": 3.4631993770599365,
|
2320 |
+
"learning_rate": 8.41154290817141e-05,
|
2321 |
+
"loss": 5.3395,
|
2322 |
+
"step": 6540
|
2323 |
+
},
|
2324 |
+
{
|
2325 |
+
"epoch": 3.6936936936936937,
|
2326 |
+
"grad_norm": 3.5585803985595703,
|
2327 |
+
"learning_rate": 8.340634026131597e-05,
|
2328 |
+
"loss": 5.3614,
|
2329 |
+
"step": 6560
|
2330 |
+
},
|
2331 |
+
{
|
2332 |
+
"epoch": 3.704954954954955,
|
2333 |
+
"grad_norm": 3.820502996444702,
|
2334 |
+
"learning_rate": 8.269875338725258e-05,
|
2335 |
+
"loss": 5.4271,
|
2336 |
+
"step": 6580
|
2337 |
+
},
|
2338 |
+
{
|
2339 |
+
"epoch": 3.7162162162162162,
|
2340 |
+
"grad_norm": 2.987173080444336,
|
2341 |
+
"learning_rate": 8.19926940104738e-05,
|
2342 |
+
"loss": 5.2998,
|
2343 |
+
"step": 6600
|
2344 |
+
},
|
2345 |
+
{
|
2346 |
+
"epoch": 3.7274774774774775,
|
2347 |
+
"grad_norm": 3.4731571674346924,
|
2348 |
+
"learning_rate": 8.128818762677149e-05,
|
2349 |
+
"loss": 5.2387,
|
2350 |
+
"step": 6620
|
2351 |
+
},
|
2352 |
+
{
|
2353 |
+
"epoch": 3.7387387387387387,
|
2354 |
+
"grad_norm": 4.059556007385254,
|
2355 |
+
"learning_rate": 8.058525967585897e-05,
|
2356 |
+
"loss": 5.306,
|
2357 |
+
"step": 6640
|
2358 |
+
},
|
2359 |
+
{
|
2360 |
+
"epoch": 3.75,
|
2361 |
+
"grad_norm": 3.9798636436462402,
|
2362 |
+
"learning_rate": 7.98839355404525e-05,
|
2363 |
+
"loss": 5.3417,
|
2364 |
+
"step": 6660
|
2365 |
+
},
|
2366 |
+
{
|
2367 |
+
"epoch": 3.7612612612612613,
|
2368 |
+
"grad_norm": 3.7810635566711426,
|
2369 |
+
"learning_rate": 7.918424054535446e-05,
|
2370 |
+
"loss": 5.2146,
|
2371 |
+
"step": 6680
|
2372 |
+
},
|
2373 |
+
{
|
2374 |
+
"epoch": 3.7725225225225225,
|
2375 |
+
"grad_norm": 3.5230772495269775,
|
2376 |
+
"learning_rate": 7.848619995653899e-05,
|
2377 |
+
"loss": 5.2648,
|
2378 |
+
"step": 6700
|
2379 |
+
},
|
2380 |
+
{
|
2381 |
+
"epoch": 3.7837837837837838,
|
2382 |
+
"grad_norm": 3.5359883308410645,
|
2383 |
+
"learning_rate": 7.778983898023973e-05,
|
2384 |
+
"loss": 5.2462,
|
2385 |
+
"step": 6720
|
2386 |
+
},
|
2387 |
+
{
|
2388 |
+
"epoch": 3.795045045045045,
|
2389 |
+
"grad_norm": 3.708984613418579,
|
2390 |
+
"learning_rate": 7.70951827620395e-05,
|
2391 |
+
"loss": 5.3829,
|
2392 |
+
"step": 6740
|
2393 |
+
},
|
2394 |
+
{
|
2395 |
+
"epoch": 3.8063063063063063,
|
2396 |
+
"grad_norm": 2.8333418369293213,
|
2397 |
+
"learning_rate": 7.640225638596235e-05,
|
2398 |
+
"loss": 5.1237,
|
2399 |
+
"step": 6760
|
2400 |
+
},
|
2401 |
+
{
|
2402 |
+
"epoch": 3.8175675675675675,
|
2403 |
+
"grad_norm": 3.36205792427063,
|
2404 |
+
"learning_rate": 7.571108487356774e-05,
|
2405 |
+
"loss": 5.2412,
|
2406 |
+
"step": 6780
|
2407 |
+
},
|
2408 |
+
{
|
2409 |
+
"epoch": 3.828828828828829,
|
2410 |
+
"grad_norm": 3.378696918487549,
|
2411 |
+
"learning_rate": 7.502169318304708e-05,
|
2412 |
+
"loss": 5.2637,
|
2413 |
+
"step": 6800
|
2414 |
+
},
|
2415 |
+
{
|
2416 |
+
"epoch": 3.84009009009009,
|
2417 |
+
"grad_norm": 3.482865571975708,
|
2418 |
+
"learning_rate": 7.433410620832237e-05,
|
2419 |
+
"loss": 5.3444,
|
2420 |
+
"step": 6820
|
2421 |
+
},
|
2422 |
+
{
|
2423 |
+
"epoch": 3.8513513513513513,
|
2424 |
+
"grad_norm": 3.399855852127075,
|
2425 |
+
"learning_rate": 7.364834877814738e-05,
|
2426 |
+
"loss": 5.2039,
|
2427 |
+
"step": 6840
|
2428 |
+
},
|
2429 |
+
{
|
2430 |
+
"epoch": 3.8626126126126126,
|
2431 |
+
"grad_norm": 3.70489764213562,
|
2432 |
+
"learning_rate": 7.296444565521113e-05,
|
2433 |
+
"loss": 5.268,
|
2434 |
+
"step": 6860
|
2435 |
+
},
|
2436 |
+
{
|
2437 |
+
"epoch": 3.873873873873874,
|
2438 |
+
"grad_norm": 2.9924323558807373,
|
2439 |
+
"learning_rate": 7.22824215352435e-05,
|
2440 |
+
"loss": 5.3238,
|
2441 |
+
"step": 6880
|
2442 |
+
},
|
2443 |
+
{
|
2444 |
+
"epoch": 3.885135135135135,
|
2445 |
+
"grad_norm": 3.4204506874084473,
|
2446 |
+
"learning_rate": 7.160230104612378e-05,
|
2447 |
+
"loss": 5.2653,
|
2448 |
+
"step": 6900
|
2449 |
+
},
|
2450 |
+
{
|
2451 |
+
"epoch": 3.8963963963963963,
|
2452 |
+
"grad_norm": 3.2533698081970215,
|
2453 |
+
"learning_rate": 7.092410874699097e-05,
|
2454 |
+
"loss": 5.1308,
|
2455 |
+
"step": 6920
|
2456 |
+
},
|
2457 |
+
{
|
2458 |
+
"epoch": 3.9076576576576576,
|
2459 |
+
"grad_norm": 3.4002726078033447,
|
2460 |
+
"learning_rate": 7.024786912735734e-05,
|
2461 |
+
"loss": 5.2096,
|
2462 |
+
"step": 6940
|
2463 |
+
},
|
2464 |
+
{
|
2465 |
+
"epoch": 3.918918918918919,
|
2466 |
+
"grad_norm": 3.259296417236328,
|
2467 |
+
"learning_rate": 6.957360660622392e-05,
|
2468 |
+
"loss": 5.2489,
|
2469 |
+
"step": 6960
|
2470 |
+
},
|
2471 |
+
{
|
2472 |
+
"epoch": 3.93018018018018,
|
2473 |
+
"grad_norm": 3.254307508468628,
|
2474 |
+
"learning_rate": 6.89013455311986e-05,
|
2475 |
+
"loss": 5.2024,
|
2476 |
+
"step": 6980
|
2477 |
+
},
|
2478 |
+
{
|
2479 |
+
"epoch": 3.9414414414414414,
|
2480 |
+
"grad_norm": 3.590085983276367,
|
2481 |
+
"learning_rate": 6.823111017761729e-05,
|
2482 |
+
"loss": 5.2317,
|
2483 |
+
"step": 7000
|
2484 |
+
},
|
2485 |
+
{
|
2486 |
+
"epoch": 3.9527027027027026,
|
2487 |
+
"grad_norm": 3.4750142097473145,
|
2488 |
+
"learning_rate": 6.756292474766692e-05,
|
2489 |
+
"loss": 5.1671,
|
2490 |
+
"step": 7020
|
2491 |
+
},
|
2492 |
+
{
|
2493 |
+
"epoch": 3.963963963963964,
|
2494 |
+
"grad_norm": 3.022782802581787,
|
2495 |
+
"learning_rate": 6.689681336951192e-05,
|
2496 |
+
"loss": 5.2441,
|
2497 |
+
"step": 7040
|
2498 |
+
},
|
2499 |
+
{
|
2500 |
+
"epoch": 3.975225225225225,
|
2501 |
+
"grad_norm": 3.6114349365234375,
|
2502 |
+
"learning_rate": 6.623280009642265e-05,
|
2503 |
+
"loss": 5.1392,
|
2504 |
+
"step": 7060
|
2505 |
+
},
|
2506 |
+
{
|
2507 |
+
"epoch": 3.9864864864864864,
|
2508 |
+
"grad_norm": 3.157512664794922,
|
2509 |
+
"learning_rate": 6.557090890590692e-05,
|
2510 |
+
"loss": 5.1179,
|
2511 |
+
"step": 7080
|
2512 |
+
},
|
2513 |
+
{
|
2514 |
+
"epoch": 3.9977477477477477,
|
2515 |
+
"grad_norm": 3.257357358932495,
|
2516 |
+
"learning_rate": 6.491116369884411e-05,
|
2517 |
+
"loss": 5.2692,
|
2518 |
+
"step": 7100
|
2519 |
+
},
|
2520 |
+
{
|
2521 |
+
"epoch": 4.0,
|
2522 |
+
"eval_loss": 5.425986289978027,
|
2523 |
+
"eval_runtime": 193.0246,
|
2524 |
+
"eval_samples_per_second": 42.44,
|
2525 |
+
"eval_steps_per_second": 5.305,
|
2526 |
+
"step": 7104
|
2527 |
+
},
|
2528 |
+
{
|
2529 |
+
"epoch": 4.009009009009009,
|
2530 |
+
"grad_norm": 3.169818639755249,
|
2531 |
+
"learning_rate": 6.425358829862231e-05,
|
2532 |
+
"loss": 4.9599,
|
2533 |
+
"step": 7120
|
2534 |
+
},
|
2535 |
+
{
|
2536 |
+
"epoch": 4.02027027027027,
|
2537 |
+
"grad_norm": 3.052263021469116,
|
2538 |
+
"learning_rate": 6.359820645027792e-05,
|
2539 |
+
"loss": 5.0446,
|
2540 |
+
"step": 7140
|
2541 |
+
},
|
2542 |
+
{
|
2543 |
+
"epoch": 4.031531531531532,
|
2544 |
+
"grad_norm": 3.136934518814087,
|
2545 |
+
"learning_rate": 6.294504181963806e-05,
|
2546 |
+
"loss": 5.0363,
|
2547 |
+
"step": 7160
|
2548 |
+
},
|
2549 |
+
{
|
2550 |
+
"epoch": 4.042792792792793,
|
2551 |
+
"grad_norm": 3.1823930740356445,
|
2552 |
+
"learning_rate": 6.229411799246631e-05,
|
2553 |
+
"loss": 5.0857,
|
2554 |
+
"step": 7180
|
2555 |
+
},
|
2556 |
+
{
|
2557 |
+
"epoch": 4.054054054054054,
|
2558 |
+
"grad_norm": 3.10451078414917,
|
2559 |
+
"learning_rate": 6.164545847361089e-05,
|
2560 |
+
"loss": 5.0356,
|
2561 |
+
"step": 7200
|
2562 |
+
},
|
2563 |
+
{
|
2564 |
+
"epoch": 4.065315315315315,
|
2565 |
+
"grad_norm": 3.185702085494995,
|
2566 |
+
"learning_rate": 6.0999086686155754e-05,
|
2567 |
+
"loss": 5.1337,
|
2568 |
+
"step": 7220
|
2569 |
+
},
|
2570 |
+
{
|
2571 |
+
"epoch": 4.076576576576577,
|
2572 |
+
"grad_norm": 3.0306313037872314,
|
2573 |
+
"learning_rate": 6.0355025970574934e-05,
|
2574 |
+
"loss": 5.1255,
|
2575 |
+
"step": 7240
|
2576 |
+
},
|
2577 |
+
{
|
2578 |
+
"epoch": 4.087837837837838,
|
2579 |
+
"grad_norm": 3.073803424835205,
|
2580 |
+
"learning_rate": 5.971329958388981e-05,
|
2581 |
+
"loss": 5.0784,
|
2582 |
+
"step": 7260
|
2583 |
+
},
|
2584 |
+
{
|
2585 |
+
"epoch": 4.099099099099099,
|
2586 |
+
"grad_norm": 3.5534238815307617,
|
2587 |
+
"learning_rate": 5.907393069882917e-05,
|
2588 |
+
"loss": 5.1128,
|
2589 |
+
"step": 7280
|
2590 |
+
},
|
2591 |
+
{
|
2592 |
+
"epoch": 4.11036036036036,
|
2593 |
+
"grad_norm": 3.408353567123413,
|
2594 |
+
"learning_rate": 5.8436942402992355e-05,
|
2595 |
+
"loss": 5.0974,
|
2596 |
+
"step": 7300
|
2597 |
+
},
|
2598 |
+
{
|
2599 |
+
"epoch": 4.121621621621622,
|
2600 |
+
"grad_norm": 3.29302716255188,
|
2601 |
+
"learning_rate": 5.7802357698015616e-05,
|
2602 |
+
"loss": 5.0923,
|
2603 |
+
"step": 7320
|
2604 |
+
},
|
2605 |
+
{
|
2606 |
+
"epoch": 4.132882882882883,
|
2607 |
+
"grad_norm": 3.7732467651367188,
|
2608 |
+
"learning_rate": 5.717019949874192e-05,
|
2609 |
+
"loss": 5.0306,
|
2610 |
+
"step": 7340
|
2611 |
+
},
|
2612 |
+
{
|
2613 |
+
"epoch": 4.1441441441441444,
|
2614 |
+
"grad_norm": 3.2979230880737305,
|
2615 |
+
"learning_rate": 5.654049063239278e-05,
|
2616 |
+
"loss": 5.0678,
|
2617 |
+
"step": 7360
|
2618 |
+
},
|
2619 |
+
{
|
2620 |
+
"epoch": 4.155405405405405,
|
2621 |
+
"grad_norm": 3.5266690254211426,
|
2622 |
+
"learning_rate": 5.591325383774447e-05,
|
2623 |
+
"loss": 4.978,
|
2624 |
+
"step": 7380
|
2625 |
+
},
|
2626 |
+
{
|
2627 |
+
"epoch": 4.166666666666667,
|
2628 |
+
"grad_norm": 3.229464530944824,
|
2629 |
+
"learning_rate": 5.528851176430685e-05,
|
2630 |
+
"loss": 5.1394,
|
2631 |
+
"step": 7400
|
2632 |
+
},
|
2633 |
+
{
|
2634 |
+
"epoch": 4.177927927927928,
|
2635 |
+
"grad_norm": 3.034528970718384,
|
2636 |
+
"learning_rate": 5.466628697150544e-05,
|
2637 |
+
"loss": 5.1061,
|
2638 |
+
"step": 7420
|
2639 |
+
},
|
2640 |
+
{
|
2641 |
+
"epoch": 4.1891891891891895,
|
2642 |
+
"grad_norm": 3.310661554336548,
|
2643 |
+
"learning_rate": 5.4046601927866726e-05,
|
2644 |
+
"loss": 5.0375,
|
2645 |
+
"step": 7440
|
2646 |
+
},
|
2647 |
+
{
|
2648 |
+
"epoch": 4.20045045045045,
|
2649 |
+
"grad_norm": 3.658907890319824,
|
2650 |
+
"learning_rate": 5.342947901020689e-05,
|
2651 |
+
"loss": 5.102,
|
2652 |
+
"step": 7460
|
2653 |
+
},
|
2654 |
+
{
|
2655 |
+
"epoch": 4.211711711711712,
|
2656 |
+
"grad_norm": 3.3838043212890625,
|
2657 |
+
"learning_rate": 5.2814940502823875e-05,
|
2658 |
+
"loss": 5.0779,
|
2659 |
+
"step": 7480
|
2660 |
+
},
|
2661 |
+
{
|
2662 |
+
"epoch": 4.222972972972973,
|
2663 |
+
"grad_norm": 3.4222939014434814,
|
2664 |
+
"learning_rate": 5.220300859669258e-05,
|
2665 |
+
"loss": 5.0779,
|
2666 |
+
"step": 7500
|
2667 |
+
},
|
2668 |
+
{
|
2669 |
+
"epoch": 4.2342342342342345,
|
2670 |
+
"grad_norm": 2.880415678024292,
|
2671 |
+
"learning_rate": 5.159370538866352e-05,
|
2672 |
+
"loss": 5.0391,
|
2673 |
+
"step": 7520
|
2674 |
+
},
|
2675 |
+
{
|
2676 |
+
"epoch": 4.245495495495495,
|
2677 |
+
"grad_norm": 3.055744171142578,
|
2678 |
+
"learning_rate": 5.0987052880665035e-05,
|
2679 |
+
"loss": 5.0656,
|
2680 |
+
"step": 7540
|
2681 |
+
},
|
2682 |
+
{
|
2683 |
+
"epoch": 4.256756756756757,
|
2684 |
+
"grad_norm": 2.940917491912842,
|
2685 |
+
"learning_rate": 5.038307297890876e-05,
|
2686 |
+
"loss": 5.0418,
|
2687 |
+
"step": 7560
|
2688 |
+
},
|
2689 |
+
{
|
2690 |
+
"epoch": 4.268018018018018,
|
2691 |
+
"grad_norm": 2.9827332496643066,
|
2692 |
+
"learning_rate": 4.9781787493098444e-05,
|
2693 |
+
"loss": 5.0616,
|
2694 |
+
"step": 7580
|
2695 |
+
},
|
2696 |
+
{
|
2697 |
+
"epoch": 4.2792792792792795,
|
2698 |
+
"grad_norm": 2.910877227783203,
|
2699 |
+
"learning_rate": 4.918321813564265e-05,
|
2700 |
+
"loss": 5.0072,
|
2701 |
+
"step": 7600
|
2702 |
+
},
|
2703 |
+
{
|
2704 |
+
"epoch": 4.29054054054054,
|
2705 |
+
"grad_norm": 3.2261264324188232,
|
2706 |
+
"learning_rate": 4.858738652087048e-05,
|
2707 |
+
"loss": 5.0949,
|
2708 |
+
"step": 7620
|
2709 |
+
},
|
2710 |
+
{
|
2711 |
+
"epoch": 4.301801801801802,
|
2712 |
+
"grad_norm": 3.5489084720611572,
|
2713 |
+
"learning_rate": 4.799431416425132e-05,
|
2714 |
+
"loss": 5.0692,
|
2715 |
+
"step": 7640
|
2716 |
+
},
|
2717 |
+
{
|
2718 |
+
"epoch": 4.313063063063063,
|
2719 |
+
"grad_norm": 3.0110628604888916,
|
2720 |
+
"learning_rate": 4.740402248161764e-05,
|
2721 |
+
"loss": 5.0054,
|
2722 |
+
"step": 7660
|
2723 |
+
},
|
2724 |
+
{
|
2725 |
+
"epoch": 4.324324324324325,
|
2726 |
+
"grad_norm": 3.045715808868408,
|
2727 |
+
"learning_rate": 4.6816532788392e-05,
|
2728 |
+
"loss": 5.1093,
|
2729 |
+
"step": 7680
|
2730 |
+
},
|
2731 |
+
{
|
2732 |
+
"epoch": 4.335585585585585,
|
2733 |
+
"grad_norm": 3.3725993633270264,
|
2734 |
+
"learning_rate": 4.623186629881698e-05,
|
2735 |
+
"loss": 5.0225,
|
2736 |
+
"step": 7700
|
2737 |
+
},
|
2738 |
+
{
|
2739 |
+
"epoch": 4.346846846846847,
|
2740 |
+
"grad_norm": 3.2622389793395996,
|
2741 |
+
"learning_rate": 4.565004412518948e-05,
|
2742 |
+
"loss": 4.9614,
|
2743 |
+
"step": 7720
|
2744 |
+
},
|
2745 |
+
{
|
2746 |
+
"epoch": 4.358108108108108,
|
2747 |
+
"grad_norm": 2.9401776790618896,
|
2748 |
+
"learning_rate": 4.507108727709816e-05,
|
2749 |
+
"loss": 5.1341,
|
2750 |
+
"step": 7740
|
2751 |
+
},
|
2752 |
+
{
|
2753 |
+
"epoch": 4.36936936936937,
|
2754 |
+
"grad_norm": 2.9128148555755615,
|
2755 |
+
"learning_rate": 4.4495016660664744e-05,
|
2756 |
+
"loss": 5.0439,
|
2757 |
+
"step": 7760
|
2758 |
+
},
|
2759 |
+
{
|
2760 |
+
"epoch": 4.38063063063063,
|
2761 |
+
"grad_norm": 3.063502550125122,
|
2762 |
+
"learning_rate": 4.392185307778926e-05,
|
2763 |
+
"loss": 4.9652,
|
2764 |
+
"step": 7780
|
2765 |
+
},
|
2766 |
+
{
|
2767 |
+
"epoch": 4.391891891891892,
|
2768 |
+
"grad_norm": 3.1251800060272217,
|
2769 |
+
"learning_rate": 4.335161722539885e-05,
|
2770 |
+
"loss": 4.8652,
|
2771 |
+
"step": 7800
|
2772 |
+
},
|
2773 |
+
{
|
2774 |
+
"epoch": 4.403153153153153,
|
2775 |
+
"grad_norm": 3.3288652896881104,
|
2776 |
+
"learning_rate": 4.278432969470024e-05,
|
2777 |
+
"loss": 5.0418,
|
2778 |
+
"step": 7820
|
2779 |
+
},
|
2780 |
+
{
|
2781 |
+
"epoch": 4.414414414414415,
|
2782 |
+
"grad_norm": 3.1995842456817627,
|
2783 |
+
"learning_rate": 4.222001097043633e-05,
|
2784 |
+
"loss": 5.0126,
|
2785 |
+
"step": 7840
|
2786 |
+
},
|
2787 |
+
{
|
2788 |
+
"epoch": 4.425675675675675,
|
2789 |
+
"grad_norm": 3.1318423748016357,
|
2790 |
+
"learning_rate": 4.1658681430146535e-05,
|
2791 |
+
"loss": 5.0385,
|
2792 |
+
"step": 7860
|
2793 |
+
},
|
2794 |
+
{
|
2795 |
+
"epoch": 4.436936936936937,
|
2796 |
+
"grad_norm": 2.9400112628936768,
|
2797 |
+
"learning_rate": 4.110036134343093e-05,
|
2798 |
+
"loss": 5.086,
|
2799 |
+
"step": 7880
|
2800 |
+
},
|
2801 |
+
{
|
2802 |
+
"epoch": 4.448198198198198,
|
2803 |
+
"grad_norm": 3.165642738342285,
|
2804 |
+
"learning_rate": 4.054507087121813e-05,
|
2805 |
+
"loss": 4.964,
|
2806 |
+
"step": 7900
|
2807 |
+
},
|
2808 |
+
{
|
2809 |
+
"epoch": 4.45945945945946,
|
2810 |
+
"grad_norm": 3.3058130741119385,
|
2811 |
+
"learning_rate": 3.999283006503743e-05,
|
2812 |
+
"loss": 5.0159,
|
2813 |
+
"step": 7920
|
2814 |
+
},
|
2815 |
+
{
|
2816 |
+
"epoch": 4.4707207207207205,
|
2817 |
+
"grad_norm": 3.0725340843200684,
|
2818 |
+
"learning_rate": 3.944365886629496e-05,
|
2819 |
+
"loss": 4.9717,
|
2820 |
+
"step": 7940
|
2821 |
+
},
|
2822 |
+
{
|
2823 |
+
"epoch": 4.481981981981982,
|
2824 |
+
"grad_norm": 3.178988218307495,
|
2825 |
+
"learning_rate": 3.8897577105553155e-05,
|
2826 |
+
"loss": 4.9235,
|
2827 |
+
"step": 7960
|
2828 |
+
},
|
2829 |
+
{
|
2830 |
+
"epoch": 4.493243243243243,
|
2831 |
+
"grad_norm": 2.975165367126465,
|
2832 |
+
"learning_rate": 3.835460450181491e-05,
|
2833 |
+
"loss": 5.0452,
|
2834 |
+
"step": 7980
|
2835 |
+
},
|
2836 |
+
{
|
2837 |
+
"epoch": 4.504504504504505,
|
2838 |
+
"grad_norm": 3.2761380672454834,
|
2839 |
+
"learning_rate": 3.781476066181161e-05,
|
2840 |
+
"loss": 4.9962,
|
2841 |
+
"step": 8000
|
2842 |
+
},
|
2843 |
+
{
|
2844 |
+
"epoch": 4.5157657657657655,
|
2845 |
+
"grad_norm": 2.9763007164001465,
|
2846 |
+
"learning_rate": 3.7278065079295026e-05,
|
2847 |
+
"loss": 4.9988,
|
2848 |
+
"step": 8020
|
2849 |
+
},
|
2850 |
+
{
|
2851 |
+
"epoch": 4.527027027027027,
|
2852 |
+
"grad_norm": 3.1083908081054688,
|
2853 |
+
"learning_rate": 3.6744537134333355e-05,
|
2854 |
+
"loss": 4.9549,
|
2855 |
+
"step": 8040
|
2856 |
+
},
|
2857 |
+
{
|
2858 |
+
"epoch": 4.538288288288288,
|
2859 |
+
"grad_norm": 2.920250654220581,
|
2860 |
+
"learning_rate": 3.6214196092611423e-05,
|
2861 |
+
"loss": 4.9694,
|
2862 |
+
"step": 8060
|
2863 |
+
},
|
2864 |
+
{
|
2865 |
+
"epoch": 4.54954954954955,
|
2866 |
+
"grad_norm": 3.211604595184326,
|
2867 |
+
"learning_rate": 3.5687061104735115e-05,
|
2868 |
+
"loss": 4.8697,
|
2869 |
+
"step": 8080
|
2870 |
+
},
|
2871 |
+
{
|
2872 |
+
"epoch": 4.5608108108108105,
|
2873 |
+
"grad_norm": 2.8755409717559814,
|
2874 |
+
"learning_rate": 3.51631512055398e-05,
|
2875 |
+
"loss": 4.9518,
|
2876 |
+
"step": 8100
|
2877 |
+
},
|
2878 |
+
{
|
2879 |
+
"epoch": 4.572072072072072,
|
2880 |
+
"grad_norm": 2.993847370147705,
|
2881 |
+
"learning_rate": 3.464248531340279e-05,
|
2882 |
+
"loss": 4.9263,
|
2883 |
+
"step": 8120
|
2884 |
+
},
|
2885 |
+
{
|
2886 |
+
"epoch": 4.583333333333333,
|
2887 |
+
"grad_norm": 3.0822536945343018,
|
2888 |
+
"learning_rate": 3.412508222956055e-05,
|
2889 |
+
"loss": 4.9402,
|
2890 |
+
"step": 8140
|
2891 |
+
},
|
2892 |
+
{
|
2893 |
+
"epoch": 4.594594594594595,
|
2894 |
+
"grad_norm": 2.793497085571289,
|
2895 |
+
"learning_rate": 3.3610960637429425e-05,
|
2896 |
+
"loss": 4.8994,
|
2897 |
+
"step": 8160
|
2898 |
+
},
|
2899 |
+
{
|
2900 |
+
"epoch": 4.6058558558558556,
|
2901 |
+
"grad_norm": 3.2247605323791504,
|
2902 |
+
"learning_rate": 3.3100139101931245e-05,
|
2903 |
+
"loss": 4.9844,
|
2904 |
+
"step": 8180
|
2905 |
+
},
|
2906 |
+
{
|
2907 |
+
"epoch": 4.617117117117117,
|
2908 |
+
"grad_norm": 2.5551505088806152,
|
2909 |
+
"learning_rate": 3.25926360688229e-05,
|
2910 |
+
"loss": 4.9373,
|
2911 |
+
"step": 8200
|
2912 |
+
},
|
2913 |
+
{
|
2914 |
+
"epoch": 4.628378378378378,
|
2915 |
+
"grad_norm": 2.971647024154663,
|
2916 |
+
"learning_rate": 3.208846986403005e-05,
|
2917 |
+
"loss": 4.8755,
|
2918 |
+
"step": 8220
|
2919 |
+
},
|
2920 |
+
{
|
2921 |
+
"epoch": 4.63963963963964,
|
2922 |
+
"grad_norm": 2.8014793395996094,
|
2923 |
+
"learning_rate": 3.158765869298574e-05,
|
2924 |
+
"loss": 4.998,
|
2925 |
+
"step": 8240
|
2926 |
+
},
|
2927 |
+
{
|
2928 |
+
"epoch": 4.650900900900901,
|
2929 |
+
"grad_norm": 2.945608615875244,
|
2930 |
+
"learning_rate": 3.109022063997262e-05,
|
2931 |
+
"loss": 5.0561,
|
2932 |
+
"step": 8260
|
2933 |
+
},
|
2934 |
+
{
|
2935 |
+
"epoch": 4.662162162162162,
|
2936 |
+
"grad_norm": 2.8917429447174072,
|
2937 |
+
"learning_rate": 3.059617366747025e-05,
|
2938 |
+
"loss": 4.9781,
|
2939 |
+
"step": 8280
|
2940 |
+
},
|
2941 |
+
{
|
2942 |
+
"epoch": 4.673423423423423,
|
2943 |
+
"grad_norm": 2.8064558506011963,
|
2944 |
+
"learning_rate": 3.0105535615506246e-05,
|
2945 |
+
"loss": 4.9577,
|
2946 |
+
"step": 8300
|
2947 |
+
},
|
2948 |
+
{
|
2949 |
+
"epoch": 4.684684684684685,
|
2950 |
+
"grad_norm": 2.9006826877593994,
|
2951 |
+
"learning_rate": 2.961832420101218e-05,
|
2952 |
+
"loss": 4.9261,
|
2953 |
+
"step": 8320
|
2954 |
+
},
|
2955 |
+
{
|
2956 |
+
"epoch": 4.695945945945946,
|
2957 |
+
"grad_norm": 3.117515802383423,
|
2958 |
+
"learning_rate": 2.9134557017183872e-05,
|
2959 |
+
"loss": 4.9872,
|
2960 |
+
"step": 8340
|
2961 |
+
},
|
2962 |
+
{
|
2963 |
+
"epoch": 4.707207207207207,
|
2964 |
+
"grad_norm": 2.933802843093872,
|
2965 |
+
"learning_rate": 2.865425153284594e-05,
|
2966 |
+
"loss": 4.8803,
|
2967 |
+
"step": 8360
|
2968 |
+
},
|
2969 |
+
{
|
2970 |
+
"epoch": 4.718468468468468,
|
2971 |
+
"grad_norm": 2.698777914047241,
|
2972 |
+
"learning_rate": 2.8177425091821086e-05,
|
2973 |
+
"loss": 4.9418,
|
2974 |
+
"step": 8380
|
2975 |
+
},
|
2976 |
+
{
|
2977 |
+
"epoch": 4.72972972972973,
|
2978 |
+
"grad_norm": 2.855353832244873,
|
2979 |
+
"learning_rate": 2.7704094912303912e-05,
|
2980 |
+
"loss": 4.8607,
|
2981 |
+
"step": 8400
|
2982 |
+
},
|
2983 |
+
{
|
2984 |
+
"epoch": 4.740990990990991,
|
2985 |
+
"grad_norm": 2.8770275115966797,
|
2986 |
+
"learning_rate": 2.723427808623903e-05,
|
2987 |
+
"loss": 4.965,
|
2988 |
+
"step": 8420
|
2989 |
+
},
|
2990 |
+
{
|
2991 |
+
"epoch": 4.752252252252252,
|
2992 |
+
"grad_norm": 2.9513511657714844,
|
2993 |
+
"learning_rate": 2.6767991578703867e-05,
|
2994 |
+
"loss": 5.0047,
|
2995 |
+
"step": 8440
|
2996 |
+
},
|
2997 |
+
{
|
2998 |
+
"epoch": 4.763513513513513,
|
2999 |
+
"grad_norm": 2.800666093826294,
|
3000 |
+
"learning_rate": 2.6305252227296185e-05,
|
3001 |
+
"loss": 4.936,
|
3002 |
+
"step": 8460
|
3003 |
+
},
|
3004 |
+
{
|
3005 |
+
"epoch": 4.774774774774775,
|
3006 |
+
"grad_norm": 2.8572280406951904,
|
3007 |
+
"learning_rate": 2.5846076741526016e-05,
|
3008 |
+
"loss": 4.8764,
|
3009 |
+
"step": 8480
|
3010 |
+
},
|
3011 |
+
{
|
3012 |
+
"epoch": 4.786036036036036,
|
3013 |
+
"grad_norm": 3.1129138469696045,
|
3014 |
+
"learning_rate": 2.5390481702212172e-05,
|
3015 |
+
"loss": 4.9409,
|
3016 |
+
"step": 8500
|
3017 |
+
},
|
3018 |
+
{
|
3019 |
+
"epoch": 4.797297297297297,
|
3020 |
+
"grad_norm": 2.8832197189331055,
|
3021 |
+
"learning_rate": 2.4938483560883614e-05,
|
3022 |
+
"loss": 4.8804,
|
3023 |
+
"step": 8520
|
3024 |
+
},
|
3025 |
+
{
|
3026 |
+
"epoch": 4.808558558558558,
|
3027 |
+
"grad_norm": 2.8778629302978516,
|
3028 |
+
"learning_rate": 2.449009863918547e-05,
|
3029 |
+
"loss": 4.9717,
|
3030 |
+
"step": 8540
|
3031 |
+
},
|
3032 |
+
{
|
3033 |
+
"epoch": 4.81981981981982,
|
3034 |
+
"grad_norm": 2.717082977294922,
|
3035 |
+
"learning_rate": 2.4045343128289556e-05,
|
3036 |
+
"loss": 4.9083,
|
3037 |
+
"step": 8560
|
3038 |
+
},
|
3039 |
+
{
|
3040 |
+
"epoch": 4.831081081081081,
|
3041 |
+
"grad_norm": 3.130682945251465,
|
3042 |
+
"learning_rate": 2.3604233088309626e-05,
|
3043 |
+
"loss": 4.9114,
|
3044 |
+
"step": 8580
|
3045 |
+
},
|
3046 |
+
{
|
3047 |
+
"epoch": 4.842342342342342,
|
3048 |
+
"grad_norm": 3.13700532913208,
|
3049 |
+
"learning_rate": 2.3166784447721564e-05,
|
3050 |
+
"loss": 4.9409,
|
3051 |
+
"step": 8600
|
3052 |
+
},
|
3053 |
+
{
|
3054 |
+
"epoch": 4.853603603603604,
|
3055 |
+
"grad_norm": 2.929036855697632,
|
3056 |
+
"learning_rate": 2.2733013002788364e-05,
|
3057 |
+
"loss": 4.9474,
|
3058 |
+
"step": 8620
|
3059 |
+
},
|
3060 |
+
{
|
3061 |
+
"epoch": 4.864864864864865,
|
3062 |
+
"grad_norm": 3.0918214321136475,
|
3063 |
+
"learning_rate": 2.2302934416989315e-05,
|
3064 |
+
"loss": 4.8345,
|
3065 |
+
"step": 8640
|
3066 |
+
},
|
3067 |
+
{
|
3068 |
+
"epoch": 4.876126126126126,
|
3069 |
+
"grad_norm": 2.9362528324127197,
|
3070 |
+
"learning_rate": 2.1876564220454712e-05,
|
3071 |
+
"loss": 4.9504,
|
3072 |
+
"step": 8660
|
3073 |
+
},
|
3074 |
+
{
|
3075 |
+
"epoch": 4.887387387387387,
|
3076 |
+
"grad_norm": 3.010586738586426,
|
3077 |
+
"learning_rate": 2.1453917809405018e-05,
|
3078 |
+
"loss": 4.8896,
|
3079 |
+
"step": 8680
|
3080 |
+
},
|
3081 |
+
{
|
3082 |
+
"epoch": 4.898648648648649,
|
3083 |
+
"grad_norm": 2.653151512145996,
|
3084 |
+
"learning_rate": 2.1035010445594874e-05,
|
3085 |
+
"loss": 4.807,
|
3086 |
+
"step": 8700
|
3087 |
+
},
|
3088 |
+
{
|
3089 |
+
"epoch": 4.90990990990991,
|
3090 |
+
"grad_norm": 2.652198314666748,
|
3091 |
+
"learning_rate": 2.061985725576186e-05,
|
3092 |
+
"loss": 4.8441,
|
3093 |
+
"step": 8720
|
3094 |
+
},
|
3095 |
+
{
|
3096 |
+
"epoch": 4.921171171171171,
|
3097 |
+
"grad_norm": 2.7888987064361572,
|
3098 |
+
"learning_rate": 2.020847323108059e-05,
|
3099 |
+
"loss": 4.9651,
|
3100 |
+
"step": 8740
|
3101 |
+
},
|
3102 |
+
{
|
3103 |
+
"epoch": 4.9324324324324325,
|
3104 |
+
"grad_norm": 2.857513904571533,
|
3105 |
+
"learning_rate": 1.980087322662101e-05,
|
3106 |
+
"loss": 4.9974,
|
3107 |
+
"step": 8760
|
3108 |
+
},
|
3109 |
+
{
|
3110 |
+
"epoch": 4.943693693693694,
|
3111 |
+
"grad_norm": 2.843057632446289,
|
3112 |
+
"learning_rate": 1.9397071960812293e-05,
|
3113 |
+
"loss": 4.8412,
|
3114 |
+
"step": 8780
|
3115 |
+
},
|
3116 |
+
{
|
3117 |
+
"epoch": 4.954954954954955,
|
3118 |
+
"grad_norm": 2.9117445945739746,
|
3119 |
+
"learning_rate": 1.8997084014911125e-05,
|
3120 |
+
"loss": 4.8854,
|
3121 |
+
"step": 8800
|
3122 |
+
},
|
3123 |
+
{
|
3124 |
+
"epoch": 4.966216216216216,
|
3125 |
+
"grad_norm": 2.8624818325042725,
|
3126 |
+
"learning_rate": 1.860092383247537e-05,
|
3127 |
+
"loss": 4.9074,
|
3128 |
+
"step": 8820
|
3129 |
+
},
|
3130 |
+
{
|
3131 |
+
"epoch": 4.9774774774774775,
|
3132 |
+
"grad_norm": 2.889586925506592,
|
3133 |
+
"learning_rate": 1.8208605718842365e-05,
|
3134 |
+
"loss": 4.813,
|
3135 |
+
"step": 8840
|
3136 |
+
},
|
3137 |
+
{
|
3138 |
+
"epoch": 4.988738738738739,
|
3139 |
+
"grad_norm": 2.582094669342041,
|
3140 |
+
"learning_rate": 1.7820143840612397e-05,
|
3141 |
+
"loss": 4.874,
|
3142 |
+
"step": 8860
|
3143 |
+
},
|
3144 |
+
{
|
3145 |
+
"epoch": 5.0,
|
3146 |
+
"grad_norm": 4.356745719909668,
|
3147 |
+
"learning_rate": 1.743555222513721e-05,
|
3148 |
+
"loss": 4.924,
|
3149 |
+
"step": 8880
|
3150 |
+
},
|
3151 |
+
{
|
3152 |
+
"epoch": 5.0,
|
3153 |
+
"eval_loss": 5.185606002807617,
|
3154 |
+
"eval_runtime": 192.9629,
|
3155 |
+
"eval_samples_per_second": 42.454,
|
3156 |
+
"eval_steps_per_second": 5.307,
|
3157 |
+
"step": 8880
|
3158 |
+
},
|
3159 |
+
{
|
3160 |
+
"epoch": 5.011261261261262,
|
3161 |
+
"grad_norm": 2.8790504932403564,
|
3162 |
+
"learning_rate": 1.7054844760013344e-05,
|
3163 |
+
"loss": 4.8061,
|
3164 |
+
"step": 8900
|
3165 |
+
},
|
3166 |
+
{
|
3167 |
+
"epoch": 5.0225225225225225,
|
3168 |
+
"grad_norm": 2.651738405227661,
|
3169 |
+
"learning_rate": 1.6678035192580836e-05,
|
3170 |
+
"loss": 4.8845,
|
3171 |
+
"step": 8920
|
3172 |
+
},
|
3173 |
+
{
|
3174 |
+
"epoch": 5.033783783783784,
|
3175 |
+
"grad_norm": 2.6467983722686768,
|
3176 |
+
"learning_rate": 1.630513712942666e-05,
|
3177 |
+
"loss": 4.8352,
|
3178 |
+
"step": 8940
|
3179 |
+
},
|
3180 |
+
{
|
3181 |
+
"epoch": 5.045045045045045,
|
3182 |
+
"grad_norm": 2.8239586353302,
|
3183 |
+
"learning_rate": 1.593616403589343e-05,
|
3184 |
+
"loss": 4.8315,
|
3185 |
+
"step": 8960
|
3186 |
+
},
|
3187 |
+
{
|
3188 |
+
"epoch": 5.056306306306307,
|
3189 |
+
"grad_norm": 2.7609245777130127,
|
3190 |
+
"learning_rate": 1.557112923559313e-05,
|
3191 |
+
"loss": 4.7989,
|
3192 |
+
"step": 8980
|
3193 |
+
},
|
3194 |
+
{
|
3195 |
+
"epoch": 5.0675675675675675,
|
3196 |
+
"grad_norm": 2.8532462120056152,
|
3197 |
+
"learning_rate": 1.5210045909926155e-05,
|
3198 |
+
"loss": 4.9209,
|
3199 |
+
"step": 9000
|
3200 |
+
},
|
3201 |
+
{
|
3202 |
+
"epoch": 5.078828828828829,
|
3203 |
+
"grad_norm": 2.6088268756866455,
|
3204 |
+
"learning_rate": 1.4852927097605148e-05,
|
3205 |
+
"loss": 4.8173,
|
3206 |
+
"step": 9020
|
3207 |
+
},
|
3208 |
+
{
|
3209 |
+
"epoch": 5.09009009009009,
|
3210 |
+
"grad_norm": 2.9059317111968994,
|
3211 |
+
"learning_rate": 1.4499785694184226e-05,
|
3212 |
+
"loss": 4.8307,
|
3213 |
+
"step": 9040
|
3214 |
+
},
|
3215 |
+
{
|
3216 |
+
"epoch": 5.101351351351352,
|
3217 |
+
"grad_norm": 2.494011163711548,
|
3218 |
+
"learning_rate": 1.4150634451593398e-05,
|
3219 |
+
"loss": 4.7723,
|
3220 |
+
"step": 9060
|
3221 |
+
},
|
3222 |
+
{
|
3223 |
+
"epoch": 5.112612612612613,
|
3224 |
+
"grad_norm": 2.8410019874572754,
|
3225 |
+
"learning_rate": 1.3805485977678e-05,
|
3226 |
+
"loss": 4.7663,
|
3227 |
+
"step": 9080
|
3228 |
+
},
|
3229 |
+
{
|
3230 |
+
"epoch": 5.123873873873874,
|
3231 |
+
"grad_norm": 2.6057417392730713,
|
3232 |
+
"learning_rate": 1.3464352735743465e-05,
|
3233 |
+
"loss": 4.8229,
|
3234 |
+
"step": 9100
|
3235 |
+
},
|
3236 |
+
{
|
3237 |
+
"epoch": 5.135135135135135,
|
3238 |
+
"grad_norm": 2.7139365673065186,
|
3239 |
+
"learning_rate": 1.3127247044105228e-05,
|
3240 |
+
"loss": 4.8017,
|
3241 |
+
"step": 9120
|
3242 |
+
},
|
3243 |
+
{
|
3244 |
+
"epoch": 5.146396396396397,
|
3245 |
+
"grad_norm": 2.635087013244629,
|
3246 |
+
"learning_rate": 1.2794181075644021e-05,
|
3247 |
+
"loss": 4.8573,
|
3248 |
+
"step": 9140
|
3249 |
+
},
|
3250 |
+
{
|
3251 |
+
"epoch": 5.157657657657658,
|
3252 |
+
"grad_norm": 2.7547223567962646,
|
3253 |
+
"learning_rate": 1.2465166857366247e-05,
|
3254 |
+
"loss": 4.8131,
|
3255 |
+
"step": 9160
|
3256 |
+
},
|
3257 |
+
{
|
3258 |
+
"epoch": 5.168918918918919,
|
3259 |
+
"grad_norm": 2.743772268295288,
|
3260 |
+
"learning_rate": 1.2140216269969598e-05,
|
3261 |
+
"loss": 4.7891,
|
3262 |
+
"step": 9180
|
3263 |
+
},
|
3264 |
+
{
|
3265 |
+
"epoch": 5.18018018018018,
|
3266 |
+
"grad_norm": 2.5214784145355225,
|
3267 |
+
"learning_rate": 1.1819341047414153e-05,
|
3268 |
+
"loss": 4.823,
|
3269 |
+
"step": 9200
|
3270 |
+
},
|
3271 |
+
{
|
3272 |
+
"epoch": 5.191441441441442,
|
3273 |
+
"grad_norm": 2.7088265419006348,
|
3274 |
+
"learning_rate": 1.1502552776498726e-05,
|
3275 |
+
"loss": 4.8935,
|
3276 |
+
"step": 9220
|
3277 |
+
},
|
3278 |
+
{
|
3279 |
+
"epoch": 5.202702702702703,
|
3280 |
+
"grad_norm": 2.632638931274414,
|
3281 |
+
"learning_rate": 1.1189862896442277e-05,
|
3282 |
+
"loss": 4.7353,
|
3283 |
+
"step": 9240
|
3284 |
+
},
|
3285 |
+
{
|
3286 |
+
"epoch": 5.213963963963964,
|
3287 |
+
"grad_norm": 2.305508852005005,
|
3288 |
+
"learning_rate": 1.0881282698470954e-05,
|
3289 |
+
"loss": 4.827,
|
3290 |
+
"step": 9260
|
3291 |
+
},
|
3292 |
+
{
|
3293 |
+
"epoch": 5.225225225225225,
|
3294 |
+
"grad_norm": 2.442047119140625,
|
3295 |
+
"learning_rate": 1.0576823325410378e-05,
|
3296 |
+
"loss": 4.7164,
|
3297 |
+
"step": 9280
|
3298 |
+
},
|
3299 |
+
{
|
3300 |
+
"epoch": 5.236486486486487,
|
3301 |
+
"grad_norm": 2.4804530143737793,
|
3302 |
+
"learning_rate": 1.0276495771283298e-05,
|
3303 |
+
"loss": 4.8804,
|
3304 |
+
"step": 9300
|
3305 |
+
},
|
3306 |
+
{
|
3307 |
+
"epoch": 5.247747747747748,
|
3308 |
+
"grad_norm": 2.687025785446167,
|
3309 |
+
"learning_rate": 9.980310880912446e-06,
|
3310 |
+
"loss": 4.8512,
|
3311 |
+
"step": 9320
|
3312 |
+
},
|
3313 |
+
{
|
3314 |
+
"epoch": 5.259009009009009,
|
3315 |
+
"grad_norm": 2.6001646518707275,
|
3316 |
+
"learning_rate": 9.688279349529136e-06,
|
3317 |
+
"loss": 4.7894,
|
3318 |
+
"step": 9340
|
3319 |
+
},
|
3320 |
+
{
|
3321 |
+
"epoch": 5.27027027027027,
|
3322 |
+
"grad_norm": 2.661602735519409,
|
3323 |
+
"learning_rate": 9.400411722386912e-06,
|
3324 |
+
"loss": 4.7727,
|
3325 |
+
"step": 9360
|
3326 |
+
},
|
3327 |
+
{
|
3328 |
+
"epoch": 5.281531531531532,
|
3329 |
+
"grad_norm": 3.1882259845733643,
|
3330 |
+
"learning_rate": 9.11671839438083e-06,
|
3331 |
+
"loss": 4.7706,
|
3332 |
+
"step": 9380
|
3333 |
+
},
|
3334 |
+
{
|
3335 |
+
"epoch": 5.292792792792793,
|
3336 |
+
"grad_norm": 2.6205930709838867,
|
3337 |
+
"learning_rate": 8.837209609672034e-06,
|
3338 |
+
"loss": 4.796,
|
3339 |
+
"step": 9400
|
3340 |
+
},
|
3341 |
+
{
|
3342 |
+
"epoch": 5.304054054054054,
|
3343 |
+
"grad_norm": 2.5202319622039795,
|
3344 |
+
"learning_rate": 8.56189546131793e-06,
|
3345 |
+
"loss": 4.8147,
|
3346 |
+
"step": 9420
|
3347 |
+
},
|
3348 |
+
{
|
3349 |
+
"epoch": 5.315315315315315,
|
3350 |
+
"grad_norm": 2.323110580444336,
|
3351 |
+
"learning_rate": 8.290785890907676e-06,
|
3352 |
+
"loss": 4.7721,
|
3353 |
+
"step": 9440
|
3354 |
+
},
|
3355 |
+
{
|
3356 |
+
"epoch": 5.326576576576577,
|
3357 |
+
"grad_norm": 2.6458232402801514,
|
3358 |
+
"learning_rate": 8.023890688203138e-06,
|
3359 |
+
"loss": 4.7435,
|
3360 |
+
"step": 9460
|
3361 |
+
},
|
3362 |
+
{
|
3363 |
+
"epoch": 5.337837837837838,
|
3364 |
+
"grad_norm": 2.767462968826294,
|
3365 |
+
"learning_rate": 7.761219490785536e-06,
|
3366 |
+
"loss": 4.7834,
|
3367 |
+
"step": 9480
|
3368 |
+
},
|
3369 |
+
{
|
3370 |
+
"epoch": 5.349099099099099,
|
3371 |
+
"grad_norm": 2.790773630142212,
|
3372 |
+
"learning_rate": 7.502781783707196e-06,
|
3373 |
+
"loss": 4.7586,
|
3374 |
+
"step": 9500
|
3375 |
+
},
|
3376 |
+
{
|
3377 |
+
"epoch": 5.36036036036036,
|
3378 |
+
"grad_norm": 2.545133352279663,
|
3379 |
+
"learning_rate": 7.248586899149312e-06,
|
3380 |
+
"loss": 4.8768,
|
3381 |
+
"step": 9520
|
3382 |
+
},
|
3383 |
+
{
|
3384 |
+
"epoch": 5.371621621621622,
|
3385 |
+
"grad_norm": 2.8105554580688477,
|
3386 |
+
"learning_rate": 6.998644016084784e-06,
|
3387 |
+
"loss": 4.7286,
|
3388 |
+
"step": 9540
|
3389 |
+
},
|
3390 |
+
{
|
3391 |
+
"epoch": 5.382882882882883,
|
3392 |
+
"grad_norm": 2.626328706741333,
|
3393 |
+
"learning_rate": 6.752962159946765e-06,
|
3394 |
+
"loss": 4.8134,
|
3395 |
+
"step": 9560
|
3396 |
+
},
|
3397 |
+
{
|
3398 |
+
"epoch": 5.3941441441441444,
|
3399 |
+
"grad_norm": 2.51377010345459,
|
3400 |
+
"learning_rate": 6.511550202302832e-06,
|
3401 |
+
"loss": 4.7596,
|
3402 |
+
"step": 9580
|
3403 |
+
},
|
3404 |
+
{
|
3405 |
+
"epoch": 5.405405405405405,
|
3406 |
+
"grad_norm": 2.748156785964966,
|
3407 |
+
"learning_rate": 6.274416860534635e-06,
|
3408 |
+
"loss": 4.7094,
|
3409 |
+
"step": 9600
|
3410 |
+
},
|
3411 |
+
{
|
3412 |
+
"epoch": 5.416666666666667,
|
3413 |
+
"grad_norm": 2.787588119506836,
|
3414 |
+
"learning_rate": 6.041570697523049e-06,
|
3415 |
+
"loss": 4.8234,
|
3416 |
+
"step": 9620
|
3417 |
+
},
|
3418 |
+
{
|
3419 |
+
"epoch": 5.427927927927928,
|
3420 |
+
"grad_norm": 2.653899908065796,
|
3421 |
+
"learning_rate": 5.813020121338997e-06,
|
3422 |
+
"loss": 4.9068,
|
3423 |
+
"step": 9640
|
3424 |
+
},
|
3425 |
+
{
|
3426 |
+
"epoch": 5.4391891891891895,
|
3427 |
+
"grad_norm": 3.0841081142425537,
|
3428 |
+
"learning_rate": 5.588773384939794e-06,
|
3429 |
+
"loss": 4.7514,
|
3430 |
+
"step": 9660
|
3431 |
+
},
|
3432 |
+
{
|
3433 |
+
"epoch": 5.45045045045045,
|
3434 |
+
"grad_norm": 2.8316450119018555,
|
3435 |
+
"learning_rate": 5.368838585871283e-06,
|
3436 |
+
"loss": 4.8007,
|
3437 |
+
"step": 9680
|
3438 |
+
},
|
3439 |
+
{
|
3440 |
+
"epoch": 5.461711711711712,
|
3441 |
+
"grad_norm": 2.622342109680176,
|
3442 |
+
"learning_rate": 5.153223665975218e-06,
|
3443 |
+
"loss": 4.8918,
|
3444 |
+
"step": 9700
|
3445 |
+
},
|
3446 |
+
{
|
3447 |
+
"epoch": 5.472972972972973,
|
3448 |
+
"grad_norm": 2.6291635036468506,
|
3449 |
+
"learning_rate": 4.941936411102646e-06,
|
3450 |
+
"loss": 4.8169,
|
3451 |
+
"step": 9720
|
3452 |
+
},
|
3453 |
+
{
|
3454 |
+
"epoch": 5.4842342342342345,
|
3455 |
+
"grad_norm": 2.7928762435913086,
|
3456 |
+
"learning_rate": 4.734984450832688e-06,
|
3457 |
+
"loss": 4.84,
|
3458 |
+
"step": 9740
|
3459 |
+
},
|
3460 |
+
{
|
3461 |
+
"epoch": 5.495495495495495,
|
3462 |
+
"grad_norm": 2.4898037910461426,
|
3463 |
+
"learning_rate": 4.532375258197066e-06,
|
3464 |
+
"loss": 4.7485,
|
3465 |
+
"step": 9760
|
3466 |
+
},
|
3467 |
+
{
|
3468 |
+
"epoch": 5.506756756756757,
|
3469 |
+
"grad_norm": 2.461247444152832,
|
3470 |
+
"learning_rate": 4.3341161494102195e-06,
|
3471 |
+
"loss": 4.802,
|
3472 |
+
"step": 9780
|
3473 |
+
},
|
3474 |
+
{
|
3475 |
+
"epoch": 5.518018018018018,
|
3476 |
+
"grad_norm": 3.408018112182617,
|
3477 |
+
"learning_rate": 4.140214283605129e-06,
|
3478 |
+
"loss": 4.808,
|
3479 |
+
"step": 9800
|
3480 |
+
},
|
3481 |
+
{
|
3482 |
+
"epoch": 5.5292792792792795,
|
3483 |
+
"grad_norm": 2.723252773284912,
|
3484 |
+
"learning_rate": 3.950676662574843e-06,
|
3485 |
+
"loss": 4.8117,
|
3486 |
+
"step": 9820
|
3487 |
+
},
|
3488 |
+
{
|
3489 |
+
"epoch": 5.54054054054054,
|
3490 |
+
"grad_norm": 2.6631016731262207,
|
3491 |
+
"learning_rate": 3.765510130519567e-06,
|
3492 |
+
"loss": 4.8206,
|
3493 |
+
"step": 9840
|
3494 |
+
},
|
3495 |
+
{
|
3496 |
+
"epoch": 5.551801801801802,
|
3497 |
+
"grad_norm": 2.5921578407287598,
|
3498 |
+
"learning_rate": 3.5847213737995738e-06,
|
3499 |
+
"loss": 4.799,
|
3500 |
+
"step": 9860
|
3501 |
+
},
|
3502 |
+
{
|
3503 |
+
"epoch": 5.563063063063063,
|
3504 |
+
"grad_norm": 2.583406448364258,
|
3505 |
+
"learning_rate": 3.408316920693752e-06,
|
3506 |
+
"loss": 4.771,
|
3507 |
+
"step": 9880
|
3508 |
+
},
|
3509 |
+
{
|
3510 |
+
"epoch": 5.574324324324325,
|
3511 |
+
"grad_norm": 2.4588005542755127,
|
3512 |
+
"learning_rate": 3.2363031411638823e-06,
|
3513 |
+
"loss": 4.7028,
|
3514 |
+
"step": 9900
|
3515 |
+
},
|
3516 |
+
{
|
3517 |
+
"epoch": 5.585585585585585,
|
3518 |
+
"grad_norm": 2.5381171703338623,
|
3519 |
+
"learning_rate": 3.0686862466245274e-06,
|
3520 |
+
"loss": 4.8448,
|
3521 |
+
"step": 9920
|
3522 |
+
},
|
3523 |
+
{
|
3524 |
+
"epoch": 5.596846846846847,
|
3525 |
+
"grad_norm": 2.744335651397705,
|
3526 |
+
"learning_rate": 2.9054722897188664e-06,
|
3527 |
+
"loss": 4.7472,
|
3528 |
+
"step": 9940
|
3529 |
+
},
|
3530 |
+
{
|
3531 |
+
"epoch": 5.608108108108108,
|
3532 |
+
"grad_norm": 2.7201008796691895,
|
3533 |
+
"learning_rate": 2.746667164100089e-06,
|
3534 |
+
"loss": 4.806,
|
3535 |
+
"step": 9960
|
3536 |
+
},
|
3537 |
+
{
|
3538 |
+
"epoch": 5.61936936936937,
|
3539 |
+
"grad_norm": 2.60044264793396,
|
3540 |
+
"learning_rate": 2.592276604218513e-06,
|
3541 |
+
"loss": 4.8105,
|
3542 |
+
"step": 9980
|
3543 |
+
},
|
3544 |
+
{
|
3545 |
+
"epoch": 5.63063063063063,
|
3546 |
+
"grad_norm": 2.544846534729004,
|
3547 |
+
"learning_rate": 2.4423061851145957e-06,
|
3548 |
+
"loss": 4.7715,
|
3549 |
+
"step": 10000
|
3550 |
+
},
|
3551 |
+
{
|
3552 |
+
"epoch": 5.641891891891892,
|
3553 |
+
"grad_norm": 2.5558741092681885,
|
3554 |
+
"learning_rate": 2.296761322217594e-06,
|
3555 |
+
"loss": 4.7969,
|
3556 |
+
"step": 10020
|
3557 |
+
},
|
3558 |
+
{
|
3559 |
+
"epoch": 5.653153153153153,
|
3560 |
+
"grad_norm": 2.537177801132202,
|
3561 |
+
"learning_rate": 2.155647271149974e-06,
|
3562 |
+
"loss": 4.7813,
|
3563 |
+
"step": 10040
|
3564 |
+
},
|
3565 |
+
{
|
3566 |
+
"epoch": 5.664414414414415,
|
3567 |
+
"grad_norm": 2.7139244079589844,
|
3568 |
+
"learning_rate": 2.0189691275376842e-06,
|
3569 |
+
"loss": 4.7306,
|
3570 |
+
"step": 10060
|
3571 |
+
},
|
3572 |
+
{
|
3573 |
+
"epoch": 5.675675675675675,
|
3574 |
+
"grad_norm": 2.5433695316314697,
|
3575 |
+
"learning_rate": 1.8867318268261518e-06,
|
3576 |
+
"loss": 4.7214,
|
3577 |
+
"step": 10080
|
3578 |
+
},
|
3579 |
+
{
|
3580 |
+
"epoch": 5.686936936936937,
|
3581 |
+
"grad_norm": 2.4332406520843506,
|
3582 |
+
"learning_rate": 1.7589401441019808e-06,
|
3583 |
+
"loss": 4.7876,
|
3584 |
+
"step": 10100
|
3585 |
+
},
|
3586 |
+
{
|
3587 |
+
"epoch": 5.698198198198198,
|
3588 |
+
"grad_norm": 2.643214225769043,
|
3589 |
+
"learning_rate": 1.635598693920673e-06,
|
3590 |
+
"loss": 4.7352,
|
3591 |
+
"step": 10120
|
3592 |
+
},
|
3593 |
+
{
|
3594 |
+
"epoch": 5.70945945945946,
|
3595 |
+
"grad_norm": 2.5900988578796387,
|
3596 |
+
"learning_rate": 1.5167119301398442e-06,
|
3597 |
+
"loss": 4.8162,
|
3598 |
+
"step": 10140
|
3599 |
+
},
|
3600 |
+
{
|
3601 |
+
"epoch": 5.7207207207207205,
|
3602 |
+
"grad_norm": 2.497743606567383,
|
3603 |
+
"learning_rate": 1.4022841457584785e-06,
|
3604 |
+
"loss": 4.7861,
|
3605 |
+
"step": 10160
|
3606 |
+
},
|
3607 |
+
{
|
3608 |
+
"epoch": 5.731981981981982,
|
3609 |
+
"grad_norm": 2.8329920768737793,
|
3610 |
+
"learning_rate": 1.2923194727618854e-06,
|
3611 |
+
"loss": 4.7491,
|
3612 |
+
"step": 10180
|
3613 |
+
},
|
3614 |
+
{
|
3615 |
+
"epoch": 5.743243243243243,
|
3616 |
+
"grad_norm": 2.4208552837371826,
|
3617 |
+
"learning_rate": 1.1868218819725273e-06,
|
3618 |
+
"loss": 4.7684,
|
3619 |
+
"step": 10200
|
3620 |
+
},
|
3621 |
+
{
|
3622 |
+
"epoch": 5.754504504504505,
|
3623 |
+
"grad_norm": 2.7252206802368164,
|
3624 |
+
"learning_rate": 1.0857951829065793e-06,
|
3625 |
+
"loss": 4.7904,
|
3626 |
+
"step": 10220
|
3627 |
+
},
|
3628 |
+
{
|
3629 |
+
"epoch": 5.7657657657657655,
|
3630 |
+
"grad_norm": 2.4612338542938232,
|
3631 |
+
"learning_rate": 9.892430236363853e-07,
|
3632 |
+
"loss": 4.7682,
|
3633 |
+
"step": 10240
|
3634 |
+
},
|
3635 |
+
{
|
3636 |
+
"epoch": 5.777027027027027,
|
3637 |
+
"grad_norm": 2.621629476547241,
|
3638 |
+
"learning_rate": 8.971688906587589e-07,
|
3639 |
+
"loss": 4.7349,
|
3640 |
+
"step": 10260
|
3641 |
+
},
|
3642 |
+
{
|
3643 |
+
"epoch": 5.788288288288288,
|
3644 |
+
"grad_norm": 2.5959367752075195,
|
3645 |
+
"learning_rate": 8.095761087690695e-07,
|
3646 |
+
"loss": 4.7374,
|
3647 |
+
"step": 10280
|
3648 |
+
},
|
3649 |
+
{
|
3650 |
+
"epoch": 5.79954954954955,
|
3651 |
+
"grad_norm": 2.6641790866851807,
|
3652 |
+
"learning_rate": 7.264678409411308e-07,
|
3653 |
+
"loss": 4.7164,
|
3654 |
+
"step": 10300
|
3655 |
+
},
|
3656 |
+
{
|
3657 |
+
"epoch": 5.8108108108108105,
|
3658 |
+
"grad_norm": 2.5384955406188965,
|
3659 |
+
"learning_rate": 6.478470882130688e-07,
|
3660 |
+
"loss": 4.734,
|
3661 |
+
"step": 10320
|
3662 |
+
},
|
3663 |
+
{
|
3664 |
+
"epoch": 5.822072072072072,
|
3665 |
+
"grad_norm": 2.6770222187042236,
|
3666 |
+
"learning_rate": 5.737166895788959e-07,
|
3667 |
+
"loss": 4.7008,
|
3668 |
+
"step": 10340
|
3669 |
+
},
|
3670 |
+
{
|
3671 |
+
"epoch": 5.833333333333333,
|
3672 |
+
"grad_norm": 2.536013603210449,
|
3673 |
+
"learning_rate": 5.040793218860639e-07,
|
3674 |
+
"loss": 4.7798,
|
3675 |
+
"step": 10360
|
3676 |
+
},
|
3677 |
+
{
|
3678 |
+
"epoch": 5.844594594594595,
|
3679 |
+
"grad_norm": 2.6112325191497803,
|
3680 |
+
"learning_rate": 4.3893749973868136e-07,
|
3681 |
+
"loss": 4.7435,
|
3682 |
+
"step": 10380
|
3683 |
+
},
|
3684 |
+
{
|
3685 |
+
"epoch": 5.8558558558558556,
|
3686 |
+
"grad_norm": 2.3617124557495117,
|
3687 |
+
"learning_rate": 3.7829357540682117e-07,
|
3688 |
+
"loss": 4.7369,
|
3689 |
+
"step": 10400
|
3690 |
+
},
|
3691 |
+
{
|
3692 |
+
"epoch": 5.867117117117117,
|
3693 |
+
"grad_norm": 2.560410261154175,
|
3694 |
+
"learning_rate": 3.2214973874156174e-07,
|
3695 |
+
"loss": 4.7537,
|
3696 |
+
"step": 10420
|
3697 |
+
},
|
3698 |
+
{
|
3699 |
+
"epoch": 5.878378378378378,
|
3700 |
+
"grad_norm": 2.5535359382629395,
|
3701 |
+
"learning_rate": 2.7050801709586905e-07,
|
3702 |
+
"loss": 4.8232,
|
3703 |
+
"step": 10440
|
3704 |
+
},
|
3705 |
+
{
|
3706 |
+
"epoch": 5.88963963963964,
|
3707 |
+
"grad_norm": 2.516218423843384,
|
3708 |
+
"learning_rate": 2.2337027525140553e-07,
|
3709 |
+
"loss": 4.8043,
|
3710 |
+
"step": 10460
|
3711 |
+
},
|
3712 |
+
{
|
3713 |
+
"epoch": 5.900900900900901,
|
3714 |
+
"grad_norm": 2.555298089981079,
|
3715 |
+
"learning_rate": 1.8073821535123658e-07,
|
3716 |
+
"loss": 4.7948,
|
3717 |
+
"step": 10480
|
3718 |
+
},
|
3719 |
+
{
|
3720 |
+
"epoch": 5.912162162162162,
|
3721 |
+
"grad_norm": 2.767887592315674,
|
3722 |
+
"learning_rate": 1.4261337683829646e-07,
|
3723 |
+
"loss": 4.7963,
|
3724 |
+
"step": 10500
|
3725 |
+
},
|
3726 |
+
{
|
3727 |
+
"epoch": 5.923423423423423,
|
3728 |
+
"grad_norm": 2.55415415763855,
|
3729 |
+
"learning_rate": 1.0899713639984931e-07,
|
3730 |
+
"loss": 4.8295,
|
3731 |
+
"step": 10520
|
3732 |
+
},
|
3733 |
+
{
|
3734 |
+
"epoch": 5.934684684684685,
|
3735 |
+
"grad_norm": 2.5507781505584717,
|
3736 |
+
"learning_rate": 7.989070791776521e-08,
|
3737 |
+
"loss": 4.7257,
|
3738 |
+
"step": 10540
|
3739 |
+
},
|
3740 |
+
{
|
3741 |
+
"epoch": 5.945945945945946,
|
3742 |
+
"grad_norm": 2.439406633377075,
|
3743 |
+
"learning_rate": 5.529514242465228e-08,
|
3744 |
+
"loss": 4.8315,
|
3745 |
+
"step": 10560
|
3746 |
+
},
|
3747 |
+
{
|
3748 |
+
"epoch": 5.957207207207207,
|
3749 |
+
"grad_norm": 2.3340978622436523,
|
3750 |
+
"learning_rate": 3.5211328065970495e-08,
|
3751 |
+
"loss": 4.79,
|
3752 |
+
"step": 10580
|
3753 |
+
},
|
3754 |
+
{
|
3755 |
+
"epoch": 5.968468468468468,
|
3756 |
+
"grad_norm": 2.731367826461792,
|
3757 |
+
"learning_rate": 1.9639990067904513e-08,
|
3758 |
+
"loss": 4.7478,
|
3759 |
+
"step": 10600
|
3760 |
+
},
|
3761 |
+
{
|
3762 |
+
"epoch": 5.97972972972973,
|
3763 |
+
"grad_norm": 2.467635154724121,
|
3764 |
+
"learning_rate": 8.581690711204071e-09,
|
3765 |
+
"loss": 4.8111,
|
3766 |
+
"step": 10620
|
3767 |
+
},
|
3768 |
+
{
|
3769 |
+
"epoch": 5.990990990990991,
|
3770 |
+
"grad_norm": 2.5952091217041016,
|
3771 |
+
"learning_rate": 2.0368293108807835e-09,
|
3772 |
+
"loss": 4.847,
|
3773 |
+
"step": 10640
|
3774 |
+
},
|
3775 |
+
{
|
3776 |
+
"epoch": 6.0,
|
3777 |
+
"eval_loss": 5.124037742614746,
|
3778 |
+
"eval_runtime": 191.7484,
|
3779 |
+
"eval_samples_per_second": 42.723,
|
3780 |
+
"eval_steps_per_second": 5.34,
|
3781 |
+
"step": 10656
|
3782 |
+
}
|
3783 |
+
],
|
3784 |
+
"logging_steps": 20,
|
3785 |
+
"max_steps": 10656,
|
3786 |
+
"num_input_tokens_seen": 0,
|
3787 |
+
"num_train_epochs": 6,
|
3788 |
+
"save_steps": 500,
|
3789 |
+
"total_flos": 6565092239867904.0,
|
3790 |
+
"train_batch_size": 16,
|
3791 |
+
"trial_name": null,
|
3792 |
+
"trial_params": null
|
3793 |
+
}
|
checkpoint-10656/training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0d726ab21919a6e8fcc8d2ea845c922c50281b0863690252ce997010ca77f123
|
3 |
+
size 5176
|
config.json
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"architectures": [
|
3 |
+
"LlamaForCausalLM"
|
4 |
+
],
|
5 |
+
"attention_bias": false,
|
6 |
+
"attention_dropout": 0.0,
|
7 |
+
"bos_token_id": 1,
|
8 |
+
"eos_token_id": 2,
|
9 |
+
"hidden_act": "silu",
|
10 |
+
"hidden_size": 512,
|
11 |
+
"initializer_range": 0.02,
|
12 |
+
"intermediate_size": 1024,
|
13 |
+
"max_position_embeddings": 256,
|
14 |
+
"model_type": "llama",
|
15 |
+
"num_attention_heads": 8,
|
16 |
+
"num_hidden_layers": 16,
|
17 |
+
"num_key_value_heads": 8,
|
18 |
+
"pad_token_id": 0,
|
19 |
+
"pretraining_tp": 1,
|
20 |
+
"rms_norm_eps": 1e-06,
|
21 |
+
"rope_scaling": null,
|
22 |
+
"rope_theta": 10000.0,
|
23 |
+
"tie_word_embeddings": false,
|
24 |
+
"torch_dtype": "float32",
|
25 |
+
"transformers_version": "4.40.0",
|
26 |
+
"use_cache": true,
|
27 |
+
"vocab_size": 16000
|
28 |
+
}
|
generation_config.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"bos_token_id": 1,
|
4 |
+
"eos_token_id": 2,
|
5 |
+
"pad_token_id": 0,
|
6 |
+
"transformers_version": "4.40.0"
|
7 |
+
}
|
merges.txt
ADDED
The diff for this file is too large to render.
See raw diff
|
|
model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:466f10e3a8cb52f000f8c00350a8c6cd3b33199c74ee996676395ea8d940e2e8
|
3 |
+
size 233392032
|
special_tokens_map.json
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"bos_token": "<s>",
|
3 |
+
"eos_token": "</s>",
|
4 |
+
"pad_token": "<pad>",
|
5 |
+
"unk_token": "<|endoftext|>"
|
6 |
+
}
|
tokenizer.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
tokenizer_config.json
ADDED
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"add_prefix_space": false,
|
3 |
+
"added_tokens_decoder": {
|
4 |
+
"0": {
|
5 |
+
"content": "<pad>",
|
6 |
+
"lstrip": false,
|
7 |
+
"normalized": false,
|
8 |
+
"rstrip": false,
|
9 |
+
"single_word": false,
|
10 |
+
"special": true
|
11 |
+
},
|
12 |
+
"1": {
|
13 |
+
"content": "<s>",
|
14 |
+
"lstrip": false,
|
15 |
+
"normalized": false,
|
16 |
+
"rstrip": false,
|
17 |
+
"single_word": false,
|
18 |
+
"special": true
|
19 |
+
},
|
20 |
+
"2": {
|
21 |
+
"content": "</s>",
|
22 |
+
"lstrip": false,
|
23 |
+
"normalized": false,
|
24 |
+
"rstrip": false,
|
25 |
+
"single_word": false,
|
26 |
+
"special": true
|
27 |
+
},
|
28 |
+
"16000": {
|
29 |
+
"content": "<|endoftext|>",
|
30 |
+
"lstrip": false,
|
31 |
+
"normalized": false,
|
32 |
+
"rstrip": false,
|
33 |
+
"single_word": false,
|
34 |
+
"special": true
|
35 |
+
}
|
36 |
+
},
|
37 |
+
"bos_token": "<s>",
|
38 |
+
"clean_up_tokenization_spaces": true,
|
39 |
+
"eos_token": "</s>",
|
40 |
+
"model_max_length": 128,
|
41 |
+
"pad_token": "<pad>",
|
42 |
+
"tokenizer_class": "GPT2Tokenizer",
|
43 |
+
"unk_token": "<|endoftext|>"
|
44 |
+
}
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0d726ab21919a6e8fcc8d2ea845c922c50281b0863690252ce997010ca77f123
|
3 |
+
size 5176
|
vocab.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|