Training in progress, step 10000, checkpoint
Browse files- checkpoint-10000/config.json +31 -0
- checkpoint-10000/generation_config.json +6 -0
- checkpoint-10000/model.safetensors +3 -0
- checkpoint-10000/optimizer.pt +3 -0
- checkpoint-10000/rng_state.pth +3 -0
- checkpoint-10000/scheduler.pt +3 -0
- checkpoint-10000/special_tokens_map.json +1 -0
- checkpoint-10000/tokenizer.json +0 -0
- checkpoint-10000/tokenizer_config.json +0 -0
- checkpoint-10000/trainer_state.json +1523 -0
- checkpoint-10000/training_args.bin +3 -0
checkpoint-10000/config.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation_function": "gelu_new",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"GPT2LMHeadModel"
|
| 5 |
+
],
|
| 6 |
+
"attn_pdrop": 0.1,
|
| 7 |
+
"bos_token_id": 50256,
|
| 8 |
+
"embd_pdrop": 0.1,
|
| 9 |
+
"eos_token_id": 50256,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"layer_norm_epsilon": 1e-05,
|
| 12 |
+
"model_type": "gpt2",
|
| 13 |
+
"n_embd": 768,
|
| 14 |
+
"n_head": 12,
|
| 15 |
+
"n_inner": null,
|
| 16 |
+
"n_layer": 12,
|
| 17 |
+
"n_positions": 1024,
|
| 18 |
+
"reorder_and_upcast_attn": false,
|
| 19 |
+
"resid_pdrop": 0.1,
|
| 20 |
+
"scale_attn_by_inverse_layer_idx": false,
|
| 21 |
+
"scale_attn_weights": true,
|
| 22 |
+
"summary_activation": null,
|
| 23 |
+
"summary_first_dropout": 0.1,
|
| 24 |
+
"summary_proj_to_labels": true,
|
| 25 |
+
"summary_type": "cls_index",
|
| 26 |
+
"summary_use_proj": true,
|
| 27 |
+
"torch_dtype": "float32",
|
| 28 |
+
"transformers_version": "4.47.0.dev0",
|
| 29 |
+
"use_cache": true,
|
| 30 |
+
"vocab_size": 52000
|
| 31 |
+
}
|
checkpoint-10000/generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 50256,
|
| 4 |
+
"eos_token_id": 50256,
|
| 5 |
+
"transformers_version": "4.47.0.dev0"
|
| 6 |
+
}
|
checkpoint-10000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6d9dc9334be0e3cb98ac2ecf7f87477c0f6fd4d87de33dd758e74e2723371a8b
|
| 3 |
+
size 503128704
|
checkpoint-10000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e0f101edaa3ad96a8bbd7ac9a13ce880d9d78035b4e5bbb6c072e47c8a51bf8d
|
| 3 |
+
size 1006351290
|
checkpoint-10000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a34276a8a29f0c48d7e41083b96ab749687f5287caf6c8c7c1fdb2d9f99139c8
|
| 3 |
+
size 14244
|
checkpoint-10000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:90c311f0069114271a5dba5d72dcffc82e7a33081dd5f150aa8c9b2278c1c497
|
| 3 |
+
size 1064
|
checkpoint-10000/special_tokens_map.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{}
|
checkpoint-10000/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-10000/tokenizer_config.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-10000/trainer_state.json
ADDED
|
@@ -0,0 +1,1523 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 3.748054027557373,
|
| 3 |
+
"best_model_checkpoint": "/scratch/cl5625/exceptions/models/100M_495/checkpoint-10000",
|
| 4 |
+
"epoch": 1.0763104079216446,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 10000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.005381552039608223,
|
| 13 |
+
"grad_norm": 6.547796249389648,
|
| 14 |
+
"learning_rate": 0.0003,
|
| 15 |
+
"loss": 8.6491,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.010763104079216447,
|
| 20 |
+
"grad_norm": 5.357585906982422,
|
| 21 |
+
"learning_rate": 0.0006,
|
| 22 |
+
"loss": 6.9125,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.01614465611882467,
|
| 27 |
+
"grad_norm": 2.4043188095092773,
|
| 28 |
+
"learning_rate": 0.0005996767589699385,
|
| 29 |
+
"loss": 6.5144,
|
| 30 |
+
"step": 150
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.021526208158432893,
|
| 34 |
+
"grad_norm": 1.361128330230713,
|
| 35 |
+
"learning_rate": 0.0005993535179398771,
|
| 36 |
+
"loss": 6.2176,
|
| 37 |
+
"step": 200
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.026907760198041114,
|
| 41 |
+
"grad_norm": 1.5913020372390747,
|
| 42 |
+
"learning_rate": 0.0005990302769098158,
|
| 43 |
+
"loss": 6.0702,
|
| 44 |
+
"step": 250
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.03228931223764934,
|
| 48 |
+
"grad_norm": 1.3549662828445435,
|
| 49 |
+
"learning_rate": 0.0005987070358797543,
|
| 50 |
+
"loss": 5.983,
|
| 51 |
+
"step": 300
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.03767086427725756,
|
| 55 |
+
"grad_norm": 1.7182697057724,
|
| 56 |
+
"learning_rate": 0.0005983837948496929,
|
| 57 |
+
"loss": 5.9004,
|
| 58 |
+
"step": 350
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.04305241631686579,
|
| 62 |
+
"grad_norm": 0.9793596863746643,
|
| 63 |
+
"learning_rate": 0.0005980605538196314,
|
| 64 |
+
"loss": 5.8284,
|
| 65 |
+
"step": 400
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.048433968356474004,
|
| 69 |
+
"grad_norm": 1.1408363580703735,
|
| 70 |
+
"learning_rate": 0.0005977373127895701,
|
| 71 |
+
"loss": 5.7264,
|
| 72 |
+
"step": 450
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.05381552039608223,
|
| 76 |
+
"grad_norm": 1.4277925491333008,
|
| 77 |
+
"learning_rate": 0.0005974140717595086,
|
| 78 |
+
"loss": 5.6398,
|
| 79 |
+
"step": 500
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.05919707243569045,
|
| 83 |
+
"grad_norm": 2.2396180629730225,
|
| 84 |
+
"learning_rate": 0.0005970908307294472,
|
| 85 |
+
"loss": 5.574,
|
| 86 |
+
"step": 550
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.06457862447529868,
|
| 90 |
+
"grad_norm": 1.143998622894287,
|
| 91 |
+
"learning_rate": 0.0005967675896993858,
|
| 92 |
+
"loss": 5.5309,
|
| 93 |
+
"step": 600
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.0699601765149069,
|
| 97 |
+
"grad_norm": 1.3793718814849854,
|
| 98 |
+
"learning_rate": 0.0005964443486693243,
|
| 99 |
+
"loss": 5.4287,
|
| 100 |
+
"step": 650
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.07534172855451512,
|
| 104 |
+
"grad_norm": 0.9985382556915283,
|
| 105 |
+
"learning_rate": 0.000596121107639263,
|
| 106 |
+
"loss": 5.3816,
|
| 107 |
+
"step": 700
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.08072328059412334,
|
| 111 |
+
"grad_norm": 1.0711445808410645,
|
| 112 |
+
"learning_rate": 0.0005957978666092015,
|
| 113 |
+
"loss": 5.3184,
|
| 114 |
+
"step": 750
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.08610483263373157,
|
| 118 |
+
"grad_norm": 1.4245736598968506,
|
| 119 |
+
"learning_rate": 0.0005954746255791401,
|
| 120 |
+
"loss": 5.2668,
|
| 121 |
+
"step": 800
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.09148638467333979,
|
| 125 |
+
"grad_norm": 1.3065918684005737,
|
| 126 |
+
"learning_rate": 0.0005951513845490787,
|
| 127 |
+
"loss": 5.2295,
|
| 128 |
+
"step": 850
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.09686793671294801,
|
| 132 |
+
"grad_norm": 1.2478379011154175,
|
| 133 |
+
"learning_rate": 0.0005948281435190174,
|
| 134 |
+
"loss": 5.1634,
|
| 135 |
+
"step": 900
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.10224948875255624,
|
| 139 |
+
"grad_norm": 0.9259421229362488,
|
| 140 |
+
"learning_rate": 0.0005945049024889559,
|
| 141 |
+
"loss": 5.1204,
|
| 142 |
+
"step": 950
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.10763104079216446,
|
| 146 |
+
"grad_norm": 1.41254460811615,
|
| 147 |
+
"learning_rate": 0.0005941816614588944,
|
| 148 |
+
"loss": 5.0846,
|
| 149 |
+
"step": 1000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.10763104079216446,
|
| 153 |
+
"eval_accuracy": 0.22689776121811187,
|
| 154 |
+
"eval_loss": 5.023209095001221,
|
| 155 |
+
"eval_runtime": 183.3261,
|
| 156 |
+
"eval_samples_per_second": 98.246,
|
| 157 |
+
"eval_steps_per_second": 6.142,
|
| 158 |
+
"step": 1000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.11301259283177269,
|
| 162 |
+
"grad_norm": 1.1088929176330566,
|
| 163 |
+
"learning_rate": 0.000593858420428833,
|
| 164 |
+
"loss": 5.0586,
|
| 165 |
+
"step": 1050
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.1183941448713809,
|
| 169 |
+
"grad_norm": 1.2266565561294556,
|
| 170 |
+
"learning_rate": 0.0005935351793987716,
|
| 171 |
+
"loss": 5.0344,
|
| 172 |
+
"step": 1100
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 0.12377569691098914,
|
| 176 |
+
"grad_norm": 1.1834657192230225,
|
| 177 |
+
"learning_rate": 0.0005932119383687103,
|
| 178 |
+
"loss": 5.0128,
|
| 179 |
+
"step": 1150
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.12915724895059735,
|
| 183 |
+
"grad_norm": 1.5358068943023682,
|
| 184 |
+
"learning_rate": 0.0005928886973386488,
|
| 185 |
+
"loss": 4.9492,
|
| 186 |
+
"step": 1200
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.13453880099020557,
|
| 190 |
+
"grad_norm": 0.8692229390144348,
|
| 191 |
+
"learning_rate": 0.0005925654563085874,
|
| 192 |
+
"loss": 4.9173,
|
| 193 |
+
"step": 1250
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.1399203530298138,
|
| 197 |
+
"grad_norm": 0.8401411175727844,
|
| 198 |
+
"learning_rate": 0.000592242215278526,
|
| 199 |
+
"loss": 4.8852,
|
| 200 |
+
"step": 1300
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.14530190506942203,
|
| 204 |
+
"grad_norm": 1.0361953973770142,
|
| 205 |
+
"learning_rate": 0.0005919189742484645,
|
| 206 |
+
"loss": 4.8926,
|
| 207 |
+
"step": 1350
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.15068345710903025,
|
| 211 |
+
"grad_norm": 0.9142393469810486,
|
| 212 |
+
"learning_rate": 0.0005915957332184032,
|
| 213 |
+
"loss": 4.8369,
|
| 214 |
+
"step": 1400
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 0.15606500914863847,
|
| 218 |
+
"grad_norm": 1.1336075067520142,
|
| 219 |
+
"learning_rate": 0.0005912724921883417,
|
| 220 |
+
"loss": 4.8169,
|
| 221 |
+
"step": 1450
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.16144656118824668,
|
| 225 |
+
"grad_norm": 1.0692909955978394,
|
| 226 |
+
"learning_rate": 0.0005909492511582803,
|
| 227 |
+
"loss": 4.8005,
|
| 228 |
+
"step": 1500
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 0.1668281132278549,
|
| 232 |
+
"grad_norm": 0.807711124420166,
|
| 233 |
+
"learning_rate": 0.0005906260101282189,
|
| 234 |
+
"loss": 4.7786,
|
| 235 |
+
"step": 1550
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.17220966526746315,
|
| 239 |
+
"grad_norm": 1.2913539409637451,
|
| 240 |
+
"learning_rate": 0.0005903027690981575,
|
| 241 |
+
"loss": 4.7333,
|
| 242 |
+
"step": 1600
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.17759121730707136,
|
| 246 |
+
"grad_norm": 1.1676956415176392,
|
| 247 |
+
"learning_rate": 0.000589979528068096,
|
| 248 |
+
"loss": 4.7357,
|
| 249 |
+
"step": 1650
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.18297276934667958,
|
| 253 |
+
"grad_norm": 0.9552448987960815,
|
| 254 |
+
"learning_rate": 0.0005896562870380347,
|
| 255 |
+
"loss": 4.7186,
|
| 256 |
+
"step": 1700
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 0.1883543213862878,
|
| 260 |
+
"grad_norm": 0.9495003819465637,
|
| 261 |
+
"learning_rate": 0.0005893330460079732,
|
| 262 |
+
"loss": 4.7037,
|
| 263 |
+
"step": 1750
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.19373587342589602,
|
| 267 |
+
"grad_norm": 0.7337987422943115,
|
| 268 |
+
"learning_rate": 0.0005890098049779118,
|
| 269 |
+
"loss": 4.6579,
|
| 270 |
+
"step": 1800
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 0.19911742546550426,
|
| 274 |
+
"grad_norm": 1.1409964561462402,
|
| 275 |
+
"learning_rate": 0.0005886865639478504,
|
| 276 |
+
"loss": 4.6561,
|
| 277 |
+
"step": 1850
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.20449897750511248,
|
| 281 |
+
"grad_norm": 0.9477244019508362,
|
| 282 |
+
"learning_rate": 0.0005883633229177889,
|
| 283 |
+
"loss": 4.6466,
|
| 284 |
+
"step": 1900
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.2098805295447207,
|
| 288 |
+
"grad_norm": 0.7901616096496582,
|
| 289 |
+
"learning_rate": 0.0005880400818877276,
|
| 290 |
+
"loss": 4.6147,
|
| 291 |
+
"step": 1950
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.2152620815843289,
|
| 295 |
+
"grad_norm": 0.9213060736656189,
|
| 296 |
+
"learning_rate": 0.0005877168408576662,
|
| 297 |
+
"loss": 4.5929,
|
| 298 |
+
"step": 2000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.2152620815843289,
|
| 302 |
+
"eval_accuracy": 0.2689594671488537,
|
| 303 |
+
"eval_loss": 4.525809288024902,
|
| 304 |
+
"eval_runtime": 181.8026,
|
| 305 |
+
"eval_samples_per_second": 99.069,
|
| 306 |
+
"eval_steps_per_second": 6.194,
|
| 307 |
+
"step": 2000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 0.22064363362393713,
|
| 311 |
+
"grad_norm": 0.9790673851966858,
|
| 312 |
+
"learning_rate": 0.0005873935998276048,
|
| 313 |
+
"loss": 4.5608,
|
| 314 |
+
"step": 2050
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.22602518566354537,
|
| 318 |
+
"grad_norm": 0.6953191161155701,
|
| 319 |
+
"learning_rate": 0.0005870703587975433,
|
| 320 |
+
"loss": 4.5888,
|
| 321 |
+
"step": 2100
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.2314067377031536,
|
| 325 |
+
"grad_norm": 0.8427320718765259,
|
| 326 |
+
"learning_rate": 0.0005867471177674818,
|
| 327 |
+
"loss": 4.548,
|
| 328 |
+
"step": 2150
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 0.2367882897427618,
|
| 332 |
+
"grad_norm": 1.2038559913635254,
|
| 333 |
+
"learning_rate": 0.0005864238767374205,
|
| 334 |
+
"loss": 4.5136,
|
| 335 |
+
"step": 2200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 0.24216984178237003,
|
| 339 |
+
"grad_norm": 0.8505734205245972,
|
| 340 |
+
"learning_rate": 0.0005861006357073591,
|
| 341 |
+
"loss": 4.5178,
|
| 342 |
+
"step": 2250
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 0.24755139382197827,
|
| 346 |
+
"grad_norm": 0.7332435846328735,
|
| 347 |
+
"learning_rate": 0.0005857773946772977,
|
| 348 |
+
"loss": 4.5166,
|
| 349 |
+
"step": 2300
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 0.2529329458615865,
|
| 353 |
+
"grad_norm": 0.9289398193359375,
|
| 354 |
+
"learning_rate": 0.0005854541536472362,
|
| 355 |
+
"loss": 4.4692,
|
| 356 |
+
"step": 2350
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.2583144979011947,
|
| 360 |
+
"grad_norm": 0.7615149617195129,
|
| 361 |
+
"learning_rate": 0.0005851309126171749,
|
| 362 |
+
"loss": 4.4647,
|
| 363 |
+
"step": 2400
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.2636960499408029,
|
| 367 |
+
"grad_norm": 1.141018033027649,
|
| 368 |
+
"learning_rate": 0.0005848076715871134,
|
| 369 |
+
"loss": 4.4292,
|
| 370 |
+
"step": 2450
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.26907760198041114,
|
| 374 |
+
"grad_norm": 0.8744039535522461,
|
| 375 |
+
"learning_rate": 0.000584484430557052,
|
| 376 |
+
"loss": 4.4525,
|
| 377 |
+
"step": 2500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.27445915402001936,
|
| 381 |
+
"grad_norm": 0.7360879778862,
|
| 382 |
+
"learning_rate": 0.0005841611895269906,
|
| 383 |
+
"loss": 4.4223,
|
| 384 |
+
"step": 2550
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.2798407060596276,
|
| 388 |
+
"grad_norm": 0.6995299458503723,
|
| 389 |
+
"learning_rate": 0.0005838379484969291,
|
| 390 |
+
"loss": 4.4034,
|
| 391 |
+
"step": 2600
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.2852222580992358,
|
| 395 |
+
"grad_norm": 0.8300929665565491,
|
| 396 |
+
"learning_rate": 0.0005835147074668678,
|
| 397 |
+
"loss": 4.4013,
|
| 398 |
+
"step": 2650
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.29060381013884407,
|
| 402 |
+
"grad_norm": 0.8035259246826172,
|
| 403 |
+
"learning_rate": 0.0005831914664368063,
|
| 404 |
+
"loss": 4.3722,
|
| 405 |
+
"step": 2700
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.2959853621784523,
|
| 409 |
+
"grad_norm": 0.8101586699485779,
|
| 410 |
+
"learning_rate": 0.0005828682254067449,
|
| 411 |
+
"loss": 4.3752,
|
| 412 |
+
"step": 2750
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 0.3013669142180605,
|
| 416 |
+
"grad_norm": 0.9427844285964966,
|
| 417 |
+
"learning_rate": 0.0005825449843766835,
|
| 418 |
+
"loss": 4.3694,
|
| 419 |
+
"step": 2800
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 0.3067484662576687,
|
| 423 |
+
"grad_norm": 0.7412756085395813,
|
| 424 |
+
"learning_rate": 0.0005822217433466221,
|
| 425 |
+
"loss": 4.3536,
|
| 426 |
+
"step": 2850
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.31213001829727693,
|
| 430 |
+
"grad_norm": 0.7558133602142334,
|
| 431 |
+
"learning_rate": 0.0005818985023165607,
|
| 432 |
+
"loss": 4.3376,
|
| 433 |
+
"step": 2900
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.31751157033688515,
|
| 437 |
+
"grad_norm": 0.8248081803321838,
|
| 438 |
+
"learning_rate": 0.0005815752612864992,
|
| 439 |
+
"loss": 4.3552,
|
| 440 |
+
"step": 2950
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 0.32289312237649337,
|
| 444 |
+
"grad_norm": 0.7637624740600586,
|
| 445 |
+
"learning_rate": 0.0005812520202564378,
|
| 446 |
+
"loss": 4.3137,
|
| 447 |
+
"step": 3000
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.32289312237649337,
|
| 451 |
+
"eval_accuracy": 0.29744379956541017,
|
| 452 |
+
"eval_loss": 4.242817401885986,
|
| 453 |
+
"eval_runtime": 181.6642,
|
| 454 |
+
"eval_samples_per_second": 99.144,
|
| 455 |
+
"eval_steps_per_second": 6.198,
|
| 456 |
+
"step": 3000
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 0.3282746744161016,
|
| 460 |
+
"grad_norm": 0.7341495156288147,
|
| 461 |
+
"learning_rate": 0.0005809287792263764,
|
| 462 |
+
"loss": 4.3128,
|
| 463 |
+
"step": 3050
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 0.3336562264557098,
|
| 467 |
+
"grad_norm": 0.7144924998283386,
|
| 468 |
+
"learning_rate": 0.0005806055381963151,
|
| 469 |
+
"loss": 4.293,
|
| 470 |
+
"step": 3100
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.3390377784953181,
|
| 474 |
+
"grad_norm": 0.7501051425933838,
|
| 475 |
+
"learning_rate": 0.0005802822971662536,
|
| 476 |
+
"loss": 4.2715,
|
| 477 |
+
"step": 3150
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.3444193305349263,
|
| 481 |
+
"grad_norm": 0.6868528723716736,
|
| 482 |
+
"learning_rate": 0.0005799590561361922,
|
| 483 |
+
"loss": 4.2802,
|
| 484 |
+
"step": 3200
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 0.3498008825745345,
|
| 488 |
+
"grad_norm": 0.8597863912582397,
|
| 489 |
+
"learning_rate": 0.0005796358151061307,
|
| 490 |
+
"loss": 4.2787,
|
| 491 |
+
"step": 3250
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.35518243461414273,
|
| 495 |
+
"grad_norm": 0.7811806201934814,
|
| 496 |
+
"learning_rate": 0.0005793125740760694,
|
| 497 |
+
"loss": 4.2557,
|
| 498 |
+
"step": 3300
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 0.36056398665375095,
|
| 502 |
+
"grad_norm": 0.641840398311615,
|
| 503 |
+
"learning_rate": 0.0005789893330460079,
|
| 504 |
+
"loss": 4.2406,
|
| 505 |
+
"step": 3350
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.36594553869335916,
|
| 509 |
+
"grad_norm": 0.860731840133667,
|
| 510 |
+
"learning_rate": 0.0005786660920159465,
|
| 511 |
+
"loss": 4.2436,
|
| 512 |
+
"step": 3400
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 0.3713270907329674,
|
| 516 |
+
"grad_norm": 0.8146591186523438,
|
| 517 |
+
"learning_rate": 0.0005783428509858851,
|
| 518 |
+
"loss": 4.2515,
|
| 519 |
+
"step": 3450
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.3767086427725756,
|
| 523 |
+
"grad_norm": 0.7766924500465393,
|
| 524 |
+
"learning_rate": 0.0005780196099558237,
|
| 525 |
+
"loss": 4.2245,
|
| 526 |
+
"step": 3500
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 0.3820901948121838,
|
| 530 |
+
"grad_norm": 0.6820563077926636,
|
| 531 |
+
"learning_rate": 0.0005776963689257623,
|
| 532 |
+
"loss": 4.2161,
|
| 533 |
+
"step": 3550
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.38747174685179203,
|
| 537 |
+
"grad_norm": 0.7251272797584534,
|
| 538 |
+
"learning_rate": 0.0005773731278957008,
|
| 539 |
+
"loss": 4.2113,
|
| 540 |
+
"step": 3600
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 0.3928532988914003,
|
| 544 |
+
"grad_norm": 0.7628197073936462,
|
| 545 |
+
"learning_rate": 0.0005770498868656394,
|
| 546 |
+
"loss": 4.2133,
|
| 547 |
+
"step": 3650
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 0.3982348509310085,
|
| 551 |
+
"grad_norm": 0.7888264656066895,
|
| 552 |
+
"learning_rate": 0.000576726645835578,
|
| 553 |
+
"loss": 4.1951,
|
| 554 |
+
"step": 3700
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.40361640297061674,
|
| 558 |
+
"grad_norm": 0.7610321640968323,
|
| 559 |
+
"learning_rate": 0.0005764034048055167,
|
| 560 |
+
"loss": 4.2038,
|
| 561 |
+
"step": 3750
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.40899795501022496,
|
| 565 |
+
"grad_norm": 0.7846929430961609,
|
| 566 |
+
"learning_rate": 0.0005760801637754552,
|
| 567 |
+
"loss": 4.1858,
|
| 568 |
+
"step": 3800
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 0.4143795070498332,
|
| 572 |
+
"grad_norm": 0.7305687665939331,
|
| 573 |
+
"learning_rate": 0.0005757569227453937,
|
| 574 |
+
"loss": 4.1908,
|
| 575 |
+
"step": 3850
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.4197610590894414,
|
| 579 |
+
"grad_norm": 0.8204297423362732,
|
| 580 |
+
"learning_rate": 0.0005754336817153324,
|
| 581 |
+
"loss": 4.1887,
|
| 582 |
+
"step": 3900
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 0.4251426111290496,
|
| 586 |
+
"grad_norm": 1.0344388484954834,
|
| 587 |
+
"learning_rate": 0.0005751104406852709,
|
| 588 |
+
"loss": 4.1589,
|
| 589 |
+
"step": 3950
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 0.4305241631686578,
|
| 593 |
+
"grad_norm": 0.7527312636375427,
|
| 594 |
+
"learning_rate": 0.0005747871996552096,
|
| 595 |
+
"loss": 4.1807,
|
| 596 |
+
"step": 4000
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 0.4305241631686578,
|
| 600 |
+
"eval_accuracy": 0.31194559184912113,
|
| 601 |
+
"eval_loss": 4.095859050750732,
|
| 602 |
+
"eval_runtime": 181.6682,
|
| 603 |
+
"eval_samples_per_second": 99.142,
|
| 604 |
+
"eval_steps_per_second": 6.198,
|
| 605 |
+
"step": 4000
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.43590571520826604,
|
| 609 |
+
"grad_norm": 0.7800742387771606,
|
| 610 |
+
"learning_rate": 0.0005744639586251481,
|
| 611 |
+
"loss": 4.1596,
|
| 612 |
+
"step": 4050
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.44128726724787426,
|
| 616 |
+
"grad_norm": 0.7501474618911743,
|
| 617 |
+
"learning_rate": 0.0005741407175950867,
|
| 618 |
+
"loss": 4.1408,
|
| 619 |
+
"step": 4100
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.44666881928748253,
|
| 623 |
+
"grad_norm": 0.7201575040817261,
|
| 624 |
+
"learning_rate": 0.0005738174765650253,
|
| 625 |
+
"loss": 4.1604,
|
| 626 |
+
"step": 4150
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.45205037132709075,
|
| 630 |
+
"grad_norm": 0.5777507424354553,
|
| 631 |
+
"learning_rate": 0.0005734942355349638,
|
| 632 |
+
"loss": 4.1428,
|
| 633 |
+
"step": 4200
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.45743192336669897,
|
| 637 |
+
"grad_norm": 0.7474133372306824,
|
| 638 |
+
"learning_rate": 0.0005731709945049025,
|
| 639 |
+
"loss": 4.1435,
|
| 640 |
+
"step": 4250
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.4628134754063072,
|
| 644 |
+
"grad_norm": 0.665733814239502,
|
| 645 |
+
"learning_rate": 0.000572847753474841,
|
| 646 |
+
"loss": 4.1553,
|
| 647 |
+
"step": 4300
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.4681950274459154,
|
| 651 |
+
"grad_norm": 0.7177342176437378,
|
| 652 |
+
"learning_rate": 0.0005725245124447796,
|
| 653 |
+
"loss": 4.1161,
|
| 654 |
+
"step": 4350
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.4735765794855236,
|
| 658 |
+
"grad_norm": 0.7497685551643372,
|
| 659 |
+
"learning_rate": 0.0005722012714147182,
|
| 660 |
+
"loss": 4.1077,
|
| 661 |
+
"step": 4400
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.47895813152513184,
|
| 665 |
+
"grad_norm": 0.7238962054252625,
|
| 666 |
+
"learning_rate": 0.0005718780303846568,
|
| 667 |
+
"loss": 4.0956,
|
| 668 |
+
"step": 4450
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.48433968356474005,
|
| 672 |
+
"grad_norm": 0.754342794418335,
|
| 673 |
+
"learning_rate": 0.0005715547893545953,
|
| 674 |
+
"loss": 4.1189,
|
| 675 |
+
"step": 4500
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.48972123560434827,
|
| 679 |
+
"grad_norm": 0.641211986541748,
|
| 680 |
+
"learning_rate": 0.000571231548324534,
|
| 681 |
+
"loss": 4.1009,
|
| 682 |
+
"step": 4550
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.49510278764395654,
|
| 686 |
+
"grad_norm": 0.902698278427124,
|
| 687 |
+
"learning_rate": 0.0005709083072944725,
|
| 688 |
+
"loss": 4.0739,
|
| 689 |
+
"step": 4600
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.5004843396835648,
|
| 693 |
+
"grad_norm": 0.6036228537559509,
|
| 694 |
+
"learning_rate": 0.0005705850662644111,
|
| 695 |
+
"loss": 4.0997,
|
| 696 |
+
"step": 4650
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.505865891723173,
|
| 700 |
+
"grad_norm": 0.6570672988891602,
|
| 701 |
+
"learning_rate": 0.0005702618252343497,
|
| 702 |
+
"loss": 4.0842,
|
| 703 |
+
"step": 4700
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.5112474437627812,
|
| 707 |
+
"grad_norm": 0.733204185962677,
|
| 708 |
+
"learning_rate": 0.0005699385842042882,
|
| 709 |
+
"loss": 4.0988,
|
| 710 |
+
"step": 4750
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.5166289958023894,
|
| 714 |
+
"grad_norm": 0.6296361088752747,
|
| 715 |
+
"learning_rate": 0.0005696153431742269,
|
| 716 |
+
"loss": 4.0862,
|
| 717 |
+
"step": 4800
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.5220105478419976,
|
| 721 |
+
"grad_norm": 0.6667340993881226,
|
| 722 |
+
"learning_rate": 0.0005692921021441655,
|
| 723 |
+
"loss": 4.0714,
|
| 724 |
+
"step": 4850
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.5273920998816058,
|
| 728 |
+
"grad_norm": 0.6166004538536072,
|
| 729 |
+
"learning_rate": 0.0005689688611141041,
|
| 730 |
+
"loss": 4.073,
|
| 731 |
+
"step": 4900
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.5327736519212141,
|
| 735 |
+
"grad_norm": 0.6546381711959839,
|
| 736 |
+
"learning_rate": 0.0005686456200840426,
|
| 737 |
+
"loss": 4.0827,
|
| 738 |
+
"step": 4950
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.5381552039608223,
|
| 742 |
+
"grad_norm": 0.6323681473731995,
|
| 743 |
+
"learning_rate": 0.0005683223790539811,
|
| 744 |
+
"loss": 4.0535,
|
| 745 |
+
"step": 5000
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.5381552039608223,
|
| 749 |
+
"eval_accuracy": 0.3216463389789736,
|
| 750 |
+
"eval_loss": 3.9916508197784424,
|
| 751 |
+
"eval_runtime": 181.8071,
|
| 752 |
+
"eval_samples_per_second": 99.067,
|
| 753 |
+
"eval_steps_per_second": 6.193,
|
| 754 |
+
"step": 5000
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 0.5435367560004305,
|
| 758 |
+
"grad_norm": 0.6683503985404968,
|
| 759 |
+
"learning_rate": 0.0005679991380239198,
|
| 760 |
+
"loss": 4.0387,
|
| 761 |
+
"step": 5050
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 0.5489183080400387,
|
| 765 |
+
"grad_norm": 0.6250083446502686,
|
| 766 |
+
"learning_rate": 0.0005676758969938584,
|
| 767 |
+
"loss": 4.0595,
|
| 768 |
+
"step": 5100
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 0.5542998600796469,
|
| 772 |
+
"grad_norm": 0.6272826194763184,
|
| 773 |
+
"learning_rate": 0.000567352655963797,
|
| 774 |
+
"loss": 4.0677,
|
| 775 |
+
"step": 5150
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 0.5596814121192552,
|
| 779 |
+
"grad_norm": 0.6267666220664978,
|
| 780 |
+
"learning_rate": 0.0005670294149337355,
|
| 781 |
+
"loss": 4.0499,
|
| 782 |
+
"step": 5200
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 0.5650629641588634,
|
| 786 |
+
"grad_norm": 0.6394859552383423,
|
| 787 |
+
"learning_rate": 0.0005667061739036742,
|
| 788 |
+
"loss": 4.0364,
|
| 789 |
+
"step": 5250
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.5704445161984716,
|
| 793 |
+
"grad_norm": 0.6560896039009094,
|
| 794 |
+
"learning_rate": 0.0005663829328736127,
|
| 795 |
+
"loss": 4.0384,
|
| 796 |
+
"step": 5300
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 0.5758260682380799,
|
| 800 |
+
"grad_norm": 0.6860178709030151,
|
| 801 |
+
"learning_rate": 0.0005660596918435512,
|
| 802 |
+
"loss": 4.0365,
|
| 803 |
+
"step": 5350
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 0.5812076202776881,
|
| 807 |
+
"grad_norm": 0.6861039400100708,
|
| 808 |
+
"learning_rate": 0.0005657364508134899,
|
| 809 |
+
"loss": 4.033,
|
| 810 |
+
"step": 5400
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 0.5865891723172963,
|
| 814 |
+
"grad_norm": 0.6428167819976807,
|
| 815 |
+
"learning_rate": 0.0005654132097834284,
|
| 816 |
+
"loss": 4.0161,
|
| 817 |
+
"step": 5450
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 0.5919707243569046,
|
| 821 |
+
"grad_norm": 0.6751614212989807,
|
| 822 |
+
"learning_rate": 0.0005650899687533671,
|
| 823 |
+
"loss": 4.0278,
|
| 824 |
+
"step": 5500
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 0.5973522763965128,
|
| 828 |
+
"grad_norm": 0.6935444474220276,
|
| 829 |
+
"learning_rate": 0.0005647667277233056,
|
| 830 |
+
"loss": 4.0194,
|
| 831 |
+
"step": 5550
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.602733828436121,
|
| 835 |
+
"grad_norm": 0.700750470161438,
|
| 836 |
+
"learning_rate": 0.0005644434866932442,
|
| 837 |
+
"loss": 4.028,
|
| 838 |
+
"step": 5600
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.6081153804757292,
|
| 842 |
+
"grad_norm": 0.641048789024353,
|
| 843 |
+
"learning_rate": 0.0005641202456631828,
|
| 844 |
+
"loss": 4.0257,
|
| 845 |
+
"step": 5650
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 0.6134969325153374,
|
| 849 |
+
"grad_norm": 0.6311193108558655,
|
| 850 |
+
"learning_rate": 0.0005637970046331214,
|
| 851 |
+
"loss": 4.005,
|
| 852 |
+
"step": 5700
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 0.6188784845549457,
|
| 856 |
+
"grad_norm": 0.7258860468864441,
|
| 857 |
+
"learning_rate": 0.00056347376360306,
|
| 858 |
+
"loss": 4.0176,
|
| 859 |
+
"step": 5750
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 0.6242600365945539,
|
| 863 |
+
"grad_norm": 0.8383513689041138,
|
| 864 |
+
"learning_rate": 0.0005631505225729985,
|
| 865 |
+
"loss": 4.0076,
|
| 866 |
+
"step": 5800
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 0.6296415886341621,
|
| 870 |
+
"grad_norm": 0.5887354612350464,
|
| 871 |
+
"learning_rate": 0.0005628272815429371,
|
| 872 |
+
"loss": 3.9825,
|
| 873 |
+
"step": 5850
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.6350231406737703,
|
| 877 |
+
"grad_norm": 0.6457569599151611,
|
| 878 |
+
"learning_rate": 0.0005625040405128757,
|
| 879 |
+
"loss": 3.9889,
|
| 880 |
+
"step": 5900
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"epoch": 0.6404046927133785,
|
| 884 |
+
"grad_norm": 0.6390018463134766,
|
| 885 |
+
"learning_rate": 0.0005621807994828143,
|
| 886 |
+
"loss": 3.9978,
|
| 887 |
+
"step": 5950
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 0.6457862447529867,
|
| 891 |
+
"grad_norm": 0.6403166651725769,
|
| 892 |
+
"learning_rate": 0.0005618575584527529,
|
| 893 |
+
"loss": 3.9888,
|
| 894 |
+
"step": 6000
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 0.6457862447529867,
|
| 898 |
+
"eval_accuracy": 0.3272947680048972,
|
| 899 |
+
"eval_loss": 3.9194672107696533,
|
| 900 |
+
"eval_runtime": 181.325,
|
| 901 |
+
"eval_samples_per_second": 99.33,
|
| 902 |
+
"eval_steps_per_second": 6.21,
|
| 903 |
+
"step": 6000
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 0.651167796792595,
|
| 907 |
+
"grad_norm": 0.6269046664237976,
|
| 908 |
+
"learning_rate": 0.0005615343174226915,
|
| 909 |
+
"loss": 3.9825,
|
| 910 |
+
"step": 6050
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 0.6565493488322032,
|
| 914 |
+
"grad_norm": 0.7226353883743286,
|
| 915 |
+
"learning_rate": 0.00056121107639263,
|
| 916 |
+
"loss": 3.9832,
|
| 917 |
+
"step": 6100
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 0.6619309008718114,
|
| 921 |
+
"grad_norm": 0.5846033692359924,
|
| 922 |
+
"learning_rate": 0.0005608878353625687,
|
| 923 |
+
"loss": 3.9839,
|
| 924 |
+
"step": 6150
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.6673124529114196,
|
| 928 |
+
"grad_norm": 0.6530674695968628,
|
| 929 |
+
"learning_rate": 0.0005605645943325072,
|
| 930 |
+
"loss": 3.9759,
|
| 931 |
+
"step": 6200
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 0.6726940049510278,
|
| 935 |
+
"grad_norm": 0.6534646153450012,
|
| 936 |
+
"learning_rate": 0.0005602413533024458,
|
| 937 |
+
"loss": 3.9751,
|
| 938 |
+
"step": 6250
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.6780755569906362,
|
| 942 |
+
"grad_norm": 0.6664760112762451,
|
| 943 |
+
"learning_rate": 0.0005599181122723844,
|
| 944 |
+
"loss": 3.9406,
|
| 945 |
+
"step": 6300
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 0.6834571090302444,
|
| 949 |
+
"grad_norm": 0.6543918251991272,
|
| 950 |
+
"learning_rate": 0.000559594871242323,
|
| 951 |
+
"loss": 3.9741,
|
| 952 |
+
"step": 6350
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 0.6888386610698526,
|
| 956 |
+
"grad_norm": 0.6636673808097839,
|
| 957 |
+
"learning_rate": 0.0005592716302122616,
|
| 958 |
+
"loss": 3.975,
|
| 959 |
+
"step": 6400
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.6942202131094608,
|
| 963 |
+
"grad_norm": 0.7524754405021667,
|
| 964 |
+
"learning_rate": 0.0005589483891822001,
|
| 965 |
+
"loss": 3.9804,
|
| 966 |
+
"step": 6450
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 0.699601765149069,
|
| 970 |
+
"grad_norm": 0.6153088808059692,
|
| 971 |
+
"learning_rate": 0.0005586251481521387,
|
| 972 |
+
"loss": 3.9682,
|
| 973 |
+
"step": 6500
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 0.7049833171886772,
|
| 977 |
+
"grad_norm": 0.62467360496521,
|
| 978 |
+
"learning_rate": 0.0005583019071220773,
|
| 979 |
+
"loss": 3.9476,
|
| 980 |
+
"step": 6550
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 0.7103648692282855,
|
| 984 |
+
"grad_norm": 0.6360256671905518,
|
| 985 |
+
"learning_rate": 0.000557978666092016,
|
| 986 |
+
"loss": 3.9448,
|
| 987 |
+
"step": 6600
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 0.7157464212678937,
|
| 991 |
+
"grad_norm": 0.5992259979248047,
|
| 992 |
+
"learning_rate": 0.0005576554250619545,
|
| 993 |
+
"loss": 3.9452,
|
| 994 |
+
"step": 6650
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 0.7211279733075019,
|
| 998 |
+
"grad_norm": 0.6049088835716248,
|
| 999 |
+
"learning_rate": 0.000557332184031893,
|
| 1000 |
+
"loss": 3.9372,
|
| 1001 |
+
"step": 6700
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 0.7265095253471101,
|
| 1005 |
+
"grad_norm": 0.6222875118255615,
|
| 1006 |
+
"learning_rate": 0.0005570089430018317,
|
| 1007 |
+
"loss": 3.95,
|
| 1008 |
+
"step": 6750
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 0.7318910773867183,
|
| 1012 |
+
"grad_norm": 0.5336934924125671,
|
| 1013 |
+
"learning_rate": 0.0005566857019717702,
|
| 1014 |
+
"loss": 3.9299,
|
| 1015 |
+
"step": 6800
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 0.7372726294263265,
|
| 1019 |
+
"grad_norm": 0.6668237447738647,
|
| 1020 |
+
"learning_rate": 0.0005563624609417089,
|
| 1021 |
+
"loss": 3.9269,
|
| 1022 |
+
"step": 6850
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 0.7426541814659348,
|
| 1026 |
+
"grad_norm": 0.5827189087867737,
|
| 1027 |
+
"learning_rate": 0.0005560392199116474,
|
| 1028 |
+
"loss": 3.934,
|
| 1029 |
+
"step": 6900
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.748035733505543,
|
| 1033 |
+
"grad_norm": 0.6519905924797058,
|
| 1034 |
+
"learning_rate": 0.000555715978881586,
|
| 1035 |
+
"loss": 3.9101,
|
| 1036 |
+
"step": 6950
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 0.7534172855451512,
|
| 1040 |
+
"grad_norm": 0.6796578764915466,
|
| 1041 |
+
"learning_rate": 0.0005553927378515246,
|
| 1042 |
+
"loss": 3.9305,
|
| 1043 |
+
"step": 7000
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 0.7534172855451512,
|
| 1047 |
+
"eval_accuracy": 0.33403450663602946,
|
| 1048 |
+
"eval_loss": 3.864957571029663,
|
| 1049 |
+
"eval_runtime": 181.9746,
|
| 1050 |
+
"eval_samples_per_second": 98.975,
|
| 1051 |
+
"eval_steps_per_second": 6.188,
|
| 1052 |
+
"step": 7000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.7587988375847594,
|
| 1056 |
+
"grad_norm": 0.561911940574646,
|
| 1057 |
+
"learning_rate": 0.0005550694968214631,
|
| 1058 |
+
"loss": 3.9368,
|
| 1059 |
+
"step": 7050
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.7641803896243676,
|
| 1063 |
+
"grad_norm": 0.6652414798736572,
|
| 1064 |
+
"learning_rate": 0.0005547462557914018,
|
| 1065 |
+
"loss": 3.9173,
|
| 1066 |
+
"step": 7100
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.7695619416639758,
|
| 1070 |
+
"grad_norm": 0.6089035868644714,
|
| 1071 |
+
"learning_rate": 0.0005544230147613403,
|
| 1072 |
+
"loss": 3.9405,
|
| 1073 |
+
"step": 7150
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.7749434937035841,
|
| 1077 |
+
"grad_norm": 0.6177041530609131,
|
| 1078 |
+
"learning_rate": 0.0005540997737312789,
|
| 1079 |
+
"loss": 3.9035,
|
| 1080 |
+
"step": 7200
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.7803250457431924,
|
| 1084 |
+
"grad_norm": 0.679482638835907,
|
| 1085 |
+
"learning_rate": 0.0005537765327012175,
|
| 1086 |
+
"loss": 3.9234,
|
| 1087 |
+
"step": 7250
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.7857065977828006,
|
| 1091 |
+
"grad_norm": 0.6637169122695923,
|
| 1092 |
+
"learning_rate": 0.0005534532916711561,
|
| 1093 |
+
"loss": 3.9177,
|
| 1094 |
+
"step": 7300
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.7910881498224088,
|
| 1098 |
+
"grad_norm": 0.6377207636833191,
|
| 1099 |
+
"learning_rate": 0.0005531300506410946,
|
| 1100 |
+
"loss": 3.9106,
|
| 1101 |
+
"step": 7350
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.796469701862017,
|
| 1105 |
+
"grad_norm": 0.6485190987586975,
|
| 1106 |
+
"learning_rate": 0.0005528068096110332,
|
| 1107 |
+
"loss": 3.9204,
|
| 1108 |
+
"step": 7400
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.8018512539016253,
|
| 1112 |
+
"grad_norm": 0.571081817150116,
|
| 1113 |
+
"learning_rate": 0.0005524835685809719,
|
| 1114 |
+
"loss": 3.8924,
|
| 1115 |
+
"step": 7450
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.8072328059412335,
|
| 1119 |
+
"grad_norm": 0.677717387676239,
|
| 1120 |
+
"learning_rate": 0.0005521603275509104,
|
| 1121 |
+
"loss": 3.9122,
|
| 1122 |
+
"step": 7500
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.8126143579808417,
|
| 1126 |
+
"grad_norm": 0.5621068477630615,
|
| 1127 |
+
"learning_rate": 0.000551837086520849,
|
| 1128 |
+
"loss": 3.8824,
|
| 1129 |
+
"step": 7550
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.8179959100204499,
|
| 1133 |
+
"grad_norm": 0.5969014167785645,
|
| 1134 |
+
"learning_rate": 0.0005515138454907875,
|
| 1135 |
+
"loss": 3.8977,
|
| 1136 |
+
"step": 7600
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.8233774620600581,
|
| 1140 |
+
"grad_norm": 0.6785698533058167,
|
| 1141 |
+
"learning_rate": 0.0005511906044607262,
|
| 1142 |
+
"loss": 3.8885,
|
| 1143 |
+
"step": 7650
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.8287590140996663,
|
| 1147 |
+
"grad_norm": 0.8015792369842529,
|
| 1148 |
+
"learning_rate": 0.0005508673634306648,
|
| 1149 |
+
"loss": 3.9005,
|
| 1150 |
+
"step": 7700
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.8341405661392746,
|
| 1154 |
+
"grad_norm": 0.6184539198875427,
|
| 1155 |
+
"learning_rate": 0.0005505441224006034,
|
| 1156 |
+
"loss": 3.9204,
|
| 1157 |
+
"step": 7750
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.8395221181788828,
|
| 1161 |
+
"grad_norm": 0.6219672560691833,
|
| 1162 |
+
"learning_rate": 0.0005502208813705419,
|
| 1163 |
+
"loss": 3.8908,
|
| 1164 |
+
"step": 7800
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.844903670218491,
|
| 1168 |
+
"grad_norm": 0.5572973489761353,
|
| 1169 |
+
"learning_rate": 0.0005498976403404805,
|
| 1170 |
+
"loss": 3.8861,
|
| 1171 |
+
"step": 7850
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.8502852222580992,
|
| 1175 |
+
"grad_norm": 0.6005538105964661,
|
| 1176 |
+
"learning_rate": 0.0005495743993104191,
|
| 1177 |
+
"loss": 3.8814,
|
| 1178 |
+
"step": 7900
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.8556667742977074,
|
| 1182 |
+
"grad_norm": 0.6196568012237549,
|
| 1183 |
+
"learning_rate": 0.0005492511582803576,
|
| 1184 |
+
"loss": 3.8699,
|
| 1185 |
+
"step": 7950
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.8610483263373157,
|
| 1189 |
+
"grad_norm": 0.5852003693580627,
|
| 1190 |
+
"learning_rate": 0.0005489279172502963,
|
| 1191 |
+
"loss": 3.8504,
|
| 1192 |
+
"step": 8000
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.8610483263373157,
|
| 1196 |
+
"eval_accuracy": 0.33771957797914537,
|
| 1197 |
+
"eval_loss": 3.815176248550415,
|
| 1198 |
+
"eval_runtime": 181.7741,
|
| 1199 |
+
"eval_samples_per_second": 99.085,
|
| 1200 |
+
"eval_steps_per_second": 6.195,
|
| 1201 |
+
"step": 8000
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 0.8664298783769239,
|
| 1205 |
+
"grad_norm": 0.6063625812530518,
|
| 1206 |
+
"learning_rate": 0.0005486176058614372,
|
| 1207 |
+
"loss": 3.8921,
|
| 1208 |
+
"step": 8050
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 0.8718114304165321,
|
| 1212 |
+
"grad_norm": 0.5820379257202148,
|
| 1213 |
+
"learning_rate": 0.0005482943648313759,
|
| 1214 |
+
"loss": 3.8736,
|
| 1215 |
+
"step": 8100
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 0.8771929824561403,
|
| 1219 |
+
"grad_norm": 0.6044704914093018,
|
| 1220 |
+
"learning_rate": 0.0005479711238013145,
|
| 1221 |
+
"loss": 3.8832,
|
| 1222 |
+
"step": 8150
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 0.8825745344957485,
|
| 1226 |
+
"grad_norm": 0.5778374671936035,
|
| 1227 |
+
"learning_rate": 0.0005476478827712531,
|
| 1228 |
+
"loss": 3.8815,
|
| 1229 |
+
"step": 8200
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 0.8879560865353568,
|
| 1233 |
+
"grad_norm": 0.5325333476066589,
|
| 1234 |
+
"learning_rate": 0.0005473246417411916,
|
| 1235 |
+
"loss": 3.8749,
|
| 1236 |
+
"step": 8250
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 0.8933376385749651,
|
| 1240 |
+
"grad_norm": 0.618618905544281,
|
| 1241 |
+
"learning_rate": 0.0005470014007111302,
|
| 1242 |
+
"loss": 3.8779,
|
| 1243 |
+
"step": 8300
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 0.8987191906145733,
|
| 1247 |
+
"grad_norm": 0.7097580432891846,
|
| 1248 |
+
"learning_rate": 0.0005466781596810688,
|
| 1249 |
+
"loss": 3.8664,
|
| 1250 |
+
"step": 8350
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 0.9041007426541815,
|
| 1254 |
+
"grad_norm": 0.6304284334182739,
|
| 1255 |
+
"learning_rate": 0.0005463549186510073,
|
| 1256 |
+
"loss": 3.859,
|
| 1257 |
+
"step": 8400
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 0.9094822946937897,
|
| 1261 |
+
"grad_norm": 0.5668020248413086,
|
| 1262 |
+
"learning_rate": 0.000546031677620946,
|
| 1263 |
+
"loss": 3.8624,
|
| 1264 |
+
"step": 8450
|
| 1265 |
+
},
|
| 1266 |
+
{
|
| 1267 |
+
"epoch": 0.9148638467333979,
|
| 1268 |
+
"grad_norm": 0.5498570203781128,
|
| 1269 |
+
"learning_rate": 0.0005457084365908845,
|
| 1270 |
+
"loss": 3.8593,
|
| 1271 |
+
"step": 8500
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 0.9202453987730062,
|
| 1275 |
+
"grad_norm": 0.5575296878814697,
|
| 1276 |
+
"learning_rate": 0.0005453851955608232,
|
| 1277 |
+
"loss": 3.8623,
|
| 1278 |
+
"step": 8550
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 0.9256269508126144,
|
| 1282 |
+
"grad_norm": 0.5803061127662659,
|
| 1283 |
+
"learning_rate": 0.0005450619545307617,
|
| 1284 |
+
"loss": 3.8653,
|
| 1285 |
+
"step": 8600
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 0.9310085028522226,
|
| 1289 |
+
"grad_norm": 0.6923871636390686,
|
| 1290 |
+
"learning_rate": 0.0005447387135007003,
|
| 1291 |
+
"loss": 3.8628,
|
| 1292 |
+
"step": 8650
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 0.9363900548918308,
|
| 1296 |
+
"grad_norm": 0.6441589593887329,
|
| 1297 |
+
"learning_rate": 0.0005444154724706389,
|
| 1298 |
+
"loss": 3.8624,
|
| 1299 |
+
"step": 8700
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 0.941771606931439,
|
| 1303 |
+
"grad_norm": 0.6883953809738159,
|
| 1304 |
+
"learning_rate": 0.0005440922314405775,
|
| 1305 |
+
"loss": 3.8515,
|
| 1306 |
+
"step": 8750
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.9471531589710472,
|
| 1310 |
+
"grad_norm": 0.6100162863731384,
|
| 1311 |
+
"learning_rate": 0.0005437689904105161,
|
| 1312 |
+
"loss": 3.8368,
|
| 1313 |
+
"step": 8800
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 0.9525347110106555,
|
| 1317 |
+
"grad_norm": 0.5788195133209229,
|
| 1318 |
+
"learning_rate": 0.0005434457493804546,
|
| 1319 |
+
"loss": 3.8312,
|
| 1320 |
+
"step": 8850
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 0.9579162630502637,
|
| 1324 |
+
"grad_norm": 0.5563709735870361,
|
| 1325 |
+
"learning_rate": 0.0005431225083503932,
|
| 1326 |
+
"loss": 3.8517,
|
| 1327 |
+
"step": 8900
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 0.9632978150898719,
|
| 1331 |
+
"grad_norm": 0.6155171990394592,
|
| 1332 |
+
"learning_rate": 0.0005427992673203318,
|
| 1333 |
+
"loss": 3.8583,
|
| 1334 |
+
"step": 8950
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 0.9686793671294801,
|
| 1338 |
+
"grad_norm": 0.5446226596832275,
|
| 1339 |
+
"learning_rate": 0.0005424760262902704,
|
| 1340 |
+
"loss": 3.832,
|
| 1341 |
+
"step": 9000
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 0.9686793671294801,
|
| 1345 |
+
"eval_accuracy": 0.34148950722764443,
|
| 1346 |
+
"eval_loss": 3.7837822437286377,
|
| 1347 |
+
"eval_runtime": 181.5161,
|
| 1348 |
+
"eval_samples_per_second": 99.225,
|
| 1349 |
+
"eval_steps_per_second": 6.203,
|
| 1350 |
+
"step": 9000
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 0.9740609191690883,
|
| 1354 |
+
"grad_norm": 0.6641152501106262,
|
| 1355 |
+
"learning_rate": 0.000542152785260209,
|
| 1356 |
+
"loss": 3.8364,
|
| 1357 |
+
"step": 9050
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 0.9794424712086965,
|
| 1361 |
+
"grad_norm": 0.5310914516448975,
|
| 1362 |
+
"learning_rate": 0.0005418295442301476,
|
| 1363 |
+
"loss": 3.8395,
|
| 1364 |
+
"step": 9100
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 0.9848240232483048,
|
| 1368 |
+
"grad_norm": 0.6875292062759399,
|
| 1369 |
+
"learning_rate": 0.0005415063032000861,
|
| 1370 |
+
"loss": 3.837,
|
| 1371 |
+
"step": 9150
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 0.9902055752879131,
|
| 1375 |
+
"grad_norm": 0.6055516600608826,
|
| 1376 |
+
"learning_rate": 0.0005411830621700248,
|
| 1377 |
+
"loss": 3.838,
|
| 1378 |
+
"step": 9200
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 0.9955871273275213,
|
| 1382 |
+
"grad_norm": 0.586495041847229,
|
| 1383 |
+
"learning_rate": 0.0005408598211399633,
|
| 1384 |
+
"loss": 3.8258,
|
| 1385 |
+
"step": 9250
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 1.0009686793671295,
|
| 1389 |
+
"grad_norm": 0.5521350502967834,
|
| 1390 |
+
"learning_rate": 0.0005405365801099019,
|
| 1391 |
+
"loss": 3.8056,
|
| 1392 |
+
"step": 9300
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 1.0063502314067376,
|
| 1396 |
+
"grad_norm": 0.543850302696228,
|
| 1397 |
+
"learning_rate": 0.0005402133390798405,
|
| 1398 |
+
"loss": 3.7712,
|
| 1399 |
+
"step": 9350
|
| 1400 |
+
},
|
| 1401 |
+
{
|
| 1402 |
+
"epoch": 1.011731783446346,
|
| 1403 |
+
"grad_norm": 0.6066914200782776,
|
| 1404 |
+
"learning_rate": 0.000539890098049779,
|
| 1405 |
+
"loss": 3.7626,
|
| 1406 |
+
"step": 9400
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 1.017113335485954,
|
| 1410 |
+
"grad_norm": 0.5474905967712402,
|
| 1411 |
+
"learning_rate": 0.0005395668570197177,
|
| 1412 |
+
"loss": 3.7664,
|
| 1413 |
+
"step": 9450
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 1.0224948875255624,
|
| 1417 |
+
"grad_norm": 0.6281277537345886,
|
| 1418 |
+
"learning_rate": 0.0005392436159896562,
|
| 1419 |
+
"loss": 3.7591,
|
| 1420 |
+
"step": 9500
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"epoch": 1.0278764395651705,
|
| 1424 |
+
"grad_norm": 0.5506016612052917,
|
| 1425 |
+
"learning_rate": 0.0005389203749595948,
|
| 1426 |
+
"loss": 3.7604,
|
| 1427 |
+
"step": 9550
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 1.0332579916047788,
|
| 1431 |
+
"grad_norm": 0.5733294486999512,
|
| 1432 |
+
"learning_rate": 0.0005385971339295334,
|
| 1433 |
+
"loss": 3.7693,
|
| 1434 |
+
"step": 9600
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 1.0386395436443872,
|
| 1438 |
+
"grad_norm": 0.6508961915969849,
|
| 1439 |
+
"learning_rate": 0.000538273892899472,
|
| 1440 |
+
"loss": 3.7602,
|
| 1441 |
+
"step": 9650
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 1.0440210956839953,
|
| 1445 |
+
"grad_norm": 0.8109516501426697,
|
| 1446 |
+
"learning_rate": 0.0005379506518694106,
|
| 1447 |
+
"loss": 3.7589,
|
| 1448 |
+
"step": 9700
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 1.0494026477236036,
|
| 1452 |
+
"grad_norm": 0.5636641979217529,
|
| 1453 |
+
"learning_rate": 0.0005376274108393491,
|
| 1454 |
+
"loss": 3.7685,
|
| 1455 |
+
"step": 9750
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 1.0547841997632117,
|
| 1459 |
+
"grad_norm": 0.5453770160675049,
|
| 1460 |
+
"learning_rate": 0.0005373041698092877,
|
| 1461 |
+
"loss": 3.748,
|
| 1462 |
+
"step": 9800
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 1.06016575180282,
|
| 1466 |
+
"grad_norm": 0.5409913063049316,
|
| 1467 |
+
"learning_rate": 0.0005369809287792263,
|
| 1468 |
+
"loss": 3.7585,
|
| 1469 |
+
"step": 9850
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 1.0655473038424281,
|
| 1473 |
+
"grad_norm": 0.5672662854194641,
|
| 1474 |
+
"learning_rate": 0.000536657687749165,
|
| 1475 |
+
"loss": 3.7782,
|
| 1476 |
+
"step": 9900
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 1.0709288558820365,
|
| 1480 |
+
"grad_norm": 0.6657660007476807,
|
| 1481 |
+
"learning_rate": 0.0005363344467191035,
|
| 1482 |
+
"loss": 3.7581,
|
| 1483 |
+
"step": 9950
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 1.0763104079216446,
|
| 1487 |
+
"grad_norm": 0.6501979231834412,
|
| 1488 |
+
"learning_rate": 0.000536011205689042,
|
| 1489 |
+
"loss": 3.7729,
|
| 1490 |
+
"step": 10000
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 1.0763104079216446,
|
| 1494 |
+
"eval_accuracy": 0.3445393937885969,
|
| 1495 |
+
"eval_loss": 3.748054027557373,
|
| 1496 |
+
"eval_runtime": 180.5408,
|
| 1497 |
+
"eval_samples_per_second": 99.761,
|
| 1498 |
+
"eval_steps_per_second": 6.237,
|
| 1499 |
+
"step": 10000
|
| 1500 |
+
}
|
| 1501 |
+
],
|
| 1502 |
+
"logging_steps": 50,
|
| 1503 |
+
"max_steps": 92910,
|
| 1504 |
+
"num_input_tokens_seen": 0,
|
| 1505 |
+
"num_train_epochs": 10,
|
| 1506 |
+
"save_steps": 10000,
|
| 1507 |
+
"stateful_callbacks": {
|
| 1508 |
+
"TrainerControl": {
|
| 1509 |
+
"args": {
|
| 1510 |
+
"should_epoch_stop": false,
|
| 1511 |
+
"should_evaluate": false,
|
| 1512 |
+
"should_log": false,
|
| 1513 |
+
"should_save": true,
|
| 1514 |
+
"should_training_stop": false
|
| 1515 |
+
},
|
| 1516 |
+
"attributes": {}
|
| 1517 |
+
}
|
| 1518 |
+
},
|
| 1519 |
+
"total_flos": 8.3609792151552e+16,
|
| 1520 |
+
"train_batch_size": 32,
|
| 1521 |
+
"trial_name": null,
|
| 1522 |
+
"trial_params": null
|
| 1523 |
+
}
|
checkpoint-10000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f11599afc4c0758396f0e41fe533161cd5207f21434bf81e841ef073d767ad4b
|
| 3 |
+
size 5304
|