Checkpoint at step 2000
Browse files- checkpoints/checkpoint-2000/config.json +39 -0
- checkpoints/checkpoint-2000/generation_config.json +6 -0
- checkpoints/checkpoint-2000/model-00001-of-00003.safetensors +3 -0
- checkpoints/checkpoint-2000/model-00002-of-00003.safetensors +3 -0
- checkpoints/checkpoint-2000/model-00003-of-00003.safetensors +3 -0
- checkpoints/checkpoint-2000/model.safetensors.index.json +0 -0
- checkpoints/checkpoint-2000/optimizer.pt +3 -0
- checkpoints/checkpoint-2000/rng_state.pth +3 -0
- checkpoints/checkpoint-2000/scheduler.pt +3 -0
- checkpoints/checkpoint-2000/trainer_state.json +1434 -0
- checkpoints/checkpoint-2000/training_args.bin +3 -0
- logs/events.out.tfevents.1752345702.5f638d684589.1618.0 +3 -0
- myolmoe/__pycache__/__init__.cpython-311.pyc +0 -0
- myolmoe/__pycache__/modeling_myolmoe.cpython-311.pyc +0 -0
- scripts/__pycache__/train.cpython-311.pyc +0 -0
checkpoints/checkpoint-2000/config.json
ADDED
|
@@ -0,0 +1,39 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"MyOlmoeForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_bias": false,
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"clip_qkv": null,
|
| 8 |
+
"eos_token_id": 50279,
|
| 9 |
+
"hidden_act": "silu",
|
| 10 |
+
"hidden_size": 2048,
|
| 11 |
+
"initializer_range": 0.02,
|
| 12 |
+
"intermediate_size": 1024,
|
| 13 |
+
"max_position_embeddings": 4096,
|
| 14 |
+
"max_small_expert_count": 64,
|
| 15 |
+
"model_type": "olmoe",
|
| 16 |
+
"norm_topk_prob": false,
|
| 17 |
+
"num_attention_heads": 16,
|
| 18 |
+
"num_experts": 64,
|
| 19 |
+
"num_experts_per_tok": 2,
|
| 20 |
+
"num_hidden_layers": 16,
|
| 21 |
+
"num_key_value_heads": 16,
|
| 22 |
+
"num_small_experts": 0,
|
| 23 |
+
"output_router_logits": false,
|
| 24 |
+
"pad_token_id": 1,
|
| 25 |
+
"rms_norm_eps": 1e-05,
|
| 26 |
+
"rope_scaling": null,
|
| 27 |
+
"rope_theta": 10000.0,
|
| 28 |
+
"router_aux_loss_coef": 0.01,
|
| 29 |
+
"small_expert_count": 64,
|
| 30 |
+
"small_expert_intermediate_ratio": 16,
|
| 31 |
+
"small_expert_intermediate_size": 0,
|
| 32 |
+
"small_expert_sparsity_coef": 0.1,
|
| 33 |
+
"small_expert_strategy": "constant",
|
| 34 |
+
"tie_word_embeddings": false,
|
| 35 |
+
"torch_dtype": "bfloat16",
|
| 36 |
+
"transformers_version": "4.53.1",
|
| 37 |
+
"use_cache": true,
|
| 38 |
+
"vocab_size": 50304
|
| 39 |
+
}
|
checkpoints/checkpoint-2000/generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"eos_token_id": 50279,
|
| 4 |
+
"pad_token_id": 1,
|
| 5 |
+
"transformers_version": "4.53.1"
|
| 6 |
+
}
|
checkpoints/checkpoint-2000/model-00001-of-00003.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9a1b9651bad1a045a178e22cf198d5070c94d4374f6331086e38801fe8d88ca3
|
| 3 |
+
size 4997482624
|
checkpoints/checkpoint-2000/model-00002-of-00003.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ef1a35ebe492f743ae711f5f19c26deea0d483fe6afff896c9597983c82cf489
|
| 3 |
+
size 4999439616
|
checkpoints/checkpoint-2000/model-00003-of-00003.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1fa0113b233d42187a671027d345c6700936a3a6ab2a0badff42c0f3a906665e
|
| 3 |
+
size 3892418912
|
checkpoints/checkpoint-2000/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoints/checkpoint-2000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1e3f89b1f6240c43ce31d5d498684a60ff8ffdf93c8c54d939523ec8741e1363
|
| 3 |
+
size 101356346
|
checkpoints/checkpoint-2000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2b66e3cc7c452b707ddac5caf0aa17618afb9bc1a0333600a22c4afb353f3165
|
| 3 |
+
size 14244
|
checkpoints/checkpoint-2000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:62d20a8f91207d464844e2203845ed682ecb9f8b05339a4cda2b086c9ee9421b
|
| 3 |
+
size 1064
|
checkpoints/checkpoint-2000/trainer_state.json
ADDED
|
@@ -0,0 +1,1434 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.09811316126737676,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 2000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0004905658063368838,
|
| 14 |
+
"grad_norm": 48.75,
|
| 15 |
+
"learning_rate": 1.471550032701112e-07,
|
| 16 |
+
"loss": 103.1825,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0009811316126737675,
|
| 21 |
+
"grad_norm": 49.5,
|
| 22 |
+
"learning_rate": 3.106605624591236e-07,
|
| 23 |
+
"loss": 103.9151,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.0014716974190106514,
|
| 28 |
+
"grad_norm": 47.0,
|
| 29 |
+
"learning_rate": 4.7416612164813603e-07,
|
| 30 |
+
"loss": 104.2889,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.001962263225347535,
|
| 35 |
+
"grad_norm": 40.0,
|
| 36 |
+
"learning_rate": 6.376716808371485e-07,
|
| 37 |
+
"loss": 104.1391,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.002452829031684419,
|
| 42 |
+
"grad_norm": 42.25,
|
| 43 |
+
"learning_rate": 8.011772400261609e-07,
|
| 44 |
+
"loss": 104.4931,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.0029433948380213027,
|
| 49 |
+
"grad_norm": 41.75,
|
| 50 |
+
"learning_rate": 9.646827992151733e-07,
|
| 51 |
+
"loss": 104.123,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.003433960644358187,
|
| 56 |
+
"grad_norm": 43.0,
|
| 57 |
+
"learning_rate": 1.1281883584041859e-06,
|
| 58 |
+
"loss": 103.4803,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.00392452645069507,
|
| 63 |
+
"grad_norm": 46.5,
|
| 64 |
+
"learning_rate": 1.2916939175931983e-06,
|
| 65 |
+
"loss": 103.4754,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.0044150922570319545,
|
| 70 |
+
"grad_norm": 44.5,
|
| 71 |
+
"learning_rate": 1.4551994767822106e-06,
|
| 72 |
+
"loss": 103.3514,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.004905658063368838,
|
| 77 |
+
"grad_norm": 43.0,
|
| 78 |
+
"learning_rate": 1.618705035971223e-06,
|
| 79 |
+
"loss": 103.8501,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.005396223869705722,
|
| 84 |
+
"grad_norm": 50.25,
|
| 85 |
+
"learning_rate": 1.7822105951602354e-06,
|
| 86 |
+
"loss": 102.954,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.0058867896760426055,
|
| 91 |
+
"grad_norm": 41.75,
|
| 92 |
+
"learning_rate": 1.945716154349248e-06,
|
| 93 |
+
"loss": 104.1795,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.006377355482379489,
|
| 98 |
+
"grad_norm": 40.0,
|
| 99 |
+
"learning_rate": 2.1092217135382606e-06,
|
| 100 |
+
"loss": 104.1945,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.006867921288716374,
|
| 105 |
+
"grad_norm": 42.5,
|
| 106 |
+
"learning_rate": 2.2727272727272728e-06,
|
| 107 |
+
"loss": 103.516,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.007358487095053257,
|
| 112 |
+
"grad_norm": 43.5,
|
| 113 |
+
"learning_rate": 2.4362328319162854e-06,
|
| 114 |
+
"loss": 103.7904,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.00784905290139014,
|
| 119 |
+
"grad_norm": 40.5,
|
| 120 |
+
"learning_rate": 2.5997383911052975e-06,
|
| 121 |
+
"loss": 103.0723,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.008339618707727025,
|
| 126 |
+
"grad_norm": 43.5,
|
| 127 |
+
"learning_rate": 2.76324395029431e-06,
|
| 128 |
+
"loss": 104.0213,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.008830184514063909,
|
| 133 |
+
"grad_norm": 42.5,
|
| 134 |
+
"learning_rate": 2.9267495094833227e-06,
|
| 135 |
+
"loss": 103.3656,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.009320750320400792,
|
| 140 |
+
"grad_norm": 44.5,
|
| 141 |
+
"learning_rate": 3.090255068672335e-06,
|
| 142 |
+
"loss": 103.67,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.009811316126737676,
|
| 147 |
+
"grad_norm": 44.75,
|
| 148 |
+
"learning_rate": 3.253760627861348e-06,
|
| 149 |
+
"loss": 104.2247,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.01030188193307456,
|
| 154 |
+
"grad_norm": 47.75,
|
| 155 |
+
"learning_rate": 3.41726618705036e-06,
|
| 156 |
+
"loss": 103.4264,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.010792447739411444,
|
| 161 |
+
"grad_norm": 46.5,
|
| 162 |
+
"learning_rate": 3.5807717462393727e-06,
|
| 163 |
+
"loss": 103.728,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.011283013545748328,
|
| 168 |
+
"grad_norm": 41.0,
|
| 169 |
+
"learning_rate": 3.744277305428385e-06,
|
| 170 |
+
"loss": 103.5934,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.011773579352085211,
|
| 175 |
+
"grad_norm": 46.25,
|
| 176 |
+
"learning_rate": 3.907782864617397e-06,
|
| 177 |
+
"loss": 102.8712,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.012264145158422095,
|
| 182 |
+
"grad_norm": 49.5,
|
| 183 |
+
"learning_rate": 4.07128842380641e-06,
|
| 184 |
+
"loss": 103.0208,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.012754710964758978,
|
| 189 |
+
"grad_norm": 48.75,
|
| 190 |
+
"learning_rate": 4.234793982995422e-06,
|
| 191 |
+
"loss": 103.3193,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.013245276771095863,
|
| 196 |
+
"grad_norm": 42.75,
|
| 197 |
+
"learning_rate": 4.398299542184434e-06,
|
| 198 |
+
"loss": 101.6977,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.013735842577432747,
|
| 203 |
+
"grad_norm": 44.5,
|
| 204 |
+
"learning_rate": 4.5618051013734465e-06,
|
| 205 |
+
"loss": 102.7333,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.01422640838376963,
|
| 210 |
+
"grad_norm": 46.75,
|
| 211 |
+
"learning_rate": 4.725310660562459e-06,
|
| 212 |
+
"loss": 102.2653,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.014716974190106515,
|
| 217 |
+
"grad_norm": 49.5,
|
| 218 |
+
"learning_rate": 4.888816219751472e-06,
|
| 219 |
+
"loss": 102.4468,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.015207539996443397,
|
| 224 |
+
"grad_norm": 41.25,
|
| 225 |
+
"learning_rate": 5.052321778940484e-06,
|
| 226 |
+
"loss": 101.577,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.01569810580278028,
|
| 231 |
+
"grad_norm": 46.25,
|
| 232 |
+
"learning_rate": 5.215827338129497e-06,
|
| 233 |
+
"loss": 102.1117,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.016188671609117165,
|
| 238 |
+
"grad_norm": 44.5,
|
| 239 |
+
"learning_rate": 5.3793328973185095e-06,
|
| 240 |
+
"loss": 102.6705,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.01667923741545405,
|
| 245 |
+
"grad_norm": 43.5,
|
| 246 |
+
"learning_rate": 5.542838456507521e-06,
|
| 247 |
+
"loss": 102.1149,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.017169803221790934,
|
| 252 |
+
"grad_norm": 45.0,
|
| 253 |
+
"learning_rate": 5.706344015696534e-06,
|
| 254 |
+
"loss": 102.8788,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.017660369028127818,
|
| 259 |
+
"grad_norm": 47.5,
|
| 260 |
+
"learning_rate": 5.869849574885546e-06,
|
| 261 |
+
"loss": 102.119,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.0181509348344647,
|
| 266 |
+
"grad_norm": 43.75,
|
| 267 |
+
"learning_rate": 6.033355134074559e-06,
|
| 268 |
+
"loss": 101.3416,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.018641500640801584,
|
| 273 |
+
"grad_norm": 44.5,
|
| 274 |
+
"learning_rate": 6.196860693263571e-06,
|
| 275 |
+
"loss": 101.6792,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.01913206644713847,
|
| 280 |
+
"grad_norm": 43.0,
|
| 281 |
+
"learning_rate": 6.360366252452583e-06,
|
| 282 |
+
"loss": 100.9905,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.019622632253475353,
|
| 287 |
+
"grad_norm": 55.25,
|
| 288 |
+
"learning_rate": 6.523871811641597e-06,
|
| 289 |
+
"loss": 100.5244,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.020113198059812237,
|
| 294 |
+
"grad_norm": 56.75,
|
| 295 |
+
"learning_rate": 6.687377370830608e-06,
|
| 296 |
+
"loss": 100.8768,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.02060376386614912,
|
| 301 |
+
"grad_norm": 48.75,
|
| 302 |
+
"learning_rate": 6.850882930019621e-06,
|
| 303 |
+
"loss": 101.3346,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.021094329672486003,
|
| 308 |
+
"grad_norm": 43.25,
|
| 309 |
+
"learning_rate": 7.014388489208633e-06,
|
| 310 |
+
"loss": 100.5578,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.021584895478822887,
|
| 315 |
+
"grad_norm": 44.0,
|
| 316 |
+
"learning_rate": 7.177894048397646e-06,
|
| 317 |
+
"loss": 99.5179,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.022075461285159772,
|
| 322 |
+
"grad_norm": 48.75,
|
| 323 |
+
"learning_rate": 7.341399607586659e-06,
|
| 324 |
+
"loss": 101.0797,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.022566027091496656,
|
| 329 |
+
"grad_norm": 44.5,
|
| 330 |
+
"learning_rate": 7.504905166775671e-06,
|
| 331 |
+
"loss": 99.5985,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.023056592897833537,
|
| 336 |
+
"grad_norm": 60.5,
|
| 337 |
+
"learning_rate": 7.668410725964683e-06,
|
| 338 |
+
"loss": 99.062,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.023547158704170422,
|
| 343 |
+
"grad_norm": 53.5,
|
| 344 |
+
"learning_rate": 7.831916285153695e-06,
|
| 345 |
+
"loss": 99.0438,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.024037724510507306,
|
| 350 |
+
"grad_norm": 51.0,
|
| 351 |
+
"learning_rate": 7.995421844342708e-06,
|
| 352 |
+
"loss": 98.5337,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.02452829031684419,
|
| 357 |
+
"grad_norm": 48.25,
|
| 358 |
+
"learning_rate": 8.15892740353172e-06,
|
| 359 |
+
"loss": 97.4177,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.025018856123181075,
|
| 364 |
+
"grad_norm": 55.25,
|
| 365 |
+
"learning_rate": 8.322432962720734e-06,
|
| 366 |
+
"loss": 97.3938,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.025509421929517957,
|
| 371 |
+
"grad_norm": 81.0,
|
| 372 |
+
"learning_rate": 8.485938521909745e-06,
|
| 373 |
+
"loss": 96.0514,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.02599998773585484,
|
| 378 |
+
"grad_norm": 70.5,
|
| 379 |
+
"learning_rate": 8.649444081098757e-06,
|
| 380 |
+
"loss": 96.4423,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.026490553542191726,
|
| 385 |
+
"grad_norm": 87.5,
|
| 386 |
+
"learning_rate": 8.812949640287769e-06,
|
| 387 |
+
"loss": 95.6965,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.02698111934852861,
|
| 392 |
+
"grad_norm": 113.0,
|
| 393 |
+
"learning_rate": 8.976455199476782e-06,
|
| 394 |
+
"loss": 94.126,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.027471685154865495,
|
| 399 |
+
"grad_norm": 147.0,
|
| 400 |
+
"learning_rate": 9.139960758665796e-06,
|
| 401 |
+
"loss": 92.2339,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.027962250961202376,
|
| 406 |
+
"grad_norm": 163.0,
|
| 407 |
+
"learning_rate": 9.303466317854807e-06,
|
| 408 |
+
"loss": 88.7373,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.02845281676753926,
|
| 413 |
+
"grad_norm": 274.0,
|
| 414 |
+
"learning_rate": 9.466971877043821e-06,
|
| 415 |
+
"loss": 85.5932,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.028943382573876145,
|
| 420 |
+
"grad_norm": 424.0,
|
| 421 |
+
"learning_rate": 9.630477436232833e-06,
|
| 422 |
+
"loss": 81.2647,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.02943394838021303,
|
| 427 |
+
"grad_norm": 356.0,
|
| 428 |
+
"learning_rate": 9.793982995421844e-06,
|
| 429 |
+
"loss": 79.4912,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.029924514186549914,
|
| 434 |
+
"grad_norm": 338.0,
|
| 435 |
+
"learning_rate": 9.957488554610856e-06,
|
| 436 |
+
"loss": 72.7454,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.030415079992886795,
|
| 441 |
+
"grad_norm": 398.0,
|
| 442 |
+
"learning_rate": 1.012099411379987e-05,
|
| 443 |
+
"loss": 66.9605,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.03090564579922368,
|
| 448 |
+
"grad_norm": 342.0,
|
| 449 |
+
"learning_rate": 1.0284499672988881e-05,
|
| 450 |
+
"loss": 63.5135,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.03139621160556056,
|
| 455 |
+
"grad_norm": 372.0,
|
| 456 |
+
"learning_rate": 1.0448005232177895e-05,
|
| 457 |
+
"loss": 57.5055,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.031886777411897445,
|
| 462 |
+
"grad_norm": 334.0,
|
| 463 |
+
"learning_rate": 1.0611510791366908e-05,
|
| 464 |
+
"loss": 51.0307,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.03237734321823433,
|
| 469 |
+
"grad_norm": 360.0,
|
| 470 |
+
"learning_rate": 1.077501635055592e-05,
|
| 471 |
+
"loss": 47.7223,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.032867909024571214,
|
| 476 |
+
"grad_norm": 270.0,
|
| 477 |
+
"learning_rate": 1.0938521909744932e-05,
|
| 478 |
+
"loss": 43.6193,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.0333584748309081,
|
| 483 |
+
"grad_norm": 252.0,
|
| 484 |
+
"learning_rate": 1.1102027468933943e-05,
|
| 485 |
+
"loss": 39.8644,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.03384904063724498,
|
| 490 |
+
"grad_norm": 232.0,
|
| 491 |
+
"learning_rate": 1.1265533028122957e-05,
|
| 492 |
+
"loss": 41.45,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.03433960644358187,
|
| 497 |
+
"grad_norm": 134.0,
|
| 498 |
+
"learning_rate": 1.1429038587311969e-05,
|
| 499 |
+
"loss": 34.4322,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.03483017224991875,
|
| 504 |
+
"grad_norm": 113.5,
|
| 505 |
+
"learning_rate": 1.1592544146500982e-05,
|
| 506 |
+
"loss": 39.6195,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.035320738056255636,
|
| 511 |
+
"grad_norm": 220.0,
|
| 512 |
+
"learning_rate": 1.1756049705689994e-05,
|
| 513 |
+
"loss": 35.0983,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.03581130386259252,
|
| 518 |
+
"grad_norm": 109.0,
|
| 519 |
+
"learning_rate": 1.1919555264879007e-05,
|
| 520 |
+
"loss": 34.4757,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.0363018696689294,
|
| 525 |
+
"grad_norm": 149.0,
|
| 526 |
+
"learning_rate": 1.2083060824068019e-05,
|
| 527 |
+
"loss": 34.2608,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.03679243547526628,
|
| 532 |
+
"grad_norm": 66.5,
|
| 533 |
+
"learning_rate": 1.224656638325703e-05,
|
| 534 |
+
"loss": 31.0653,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.03728300128160317,
|
| 539 |
+
"grad_norm": 96.5,
|
| 540 |
+
"learning_rate": 1.2410071942446044e-05,
|
| 541 |
+
"loss": 32.5303,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.03777356708794005,
|
| 546 |
+
"grad_norm": 82.5,
|
| 547 |
+
"learning_rate": 1.2573577501635056e-05,
|
| 548 |
+
"loss": 32.1891,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.03826413289427694,
|
| 553 |
+
"grad_norm": 72.0,
|
| 554 |
+
"learning_rate": 1.2737083060824068e-05,
|
| 555 |
+
"loss": 32.1906,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.03875469870061382,
|
| 560 |
+
"grad_norm": 58.25,
|
| 561 |
+
"learning_rate": 1.2900588620013083e-05,
|
| 562 |
+
"loss": 32.5546,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.039245264506950706,
|
| 567 |
+
"grad_norm": 141.0,
|
| 568 |
+
"learning_rate": 1.3064094179202095e-05,
|
| 569 |
+
"loss": 29.5903,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.03973583031328759,
|
| 574 |
+
"grad_norm": 87.5,
|
| 575 |
+
"learning_rate": 1.3227599738391105e-05,
|
| 576 |
+
"loss": 28.0366,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.040226396119624475,
|
| 581 |
+
"grad_norm": 89.5,
|
| 582 |
+
"learning_rate": 1.3391105297580116e-05,
|
| 583 |
+
"loss": 28.156,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.04071696192596136,
|
| 588 |
+
"grad_norm": 57.0,
|
| 589 |
+
"learning_rate": 1.3554610856769132e-05,
|
| 590 |
+
"loss": 28.1194,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.04120752773229824,
|
| 595 |
+
"grad_norm": 58.0,
|
| 596 |
+
"learning_rate": 1.3718116415958143e-05,
|
| 597 |
+
"loss": 24.1945,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.04169809353863512,
|
| 602 |
+
"grad_norm": 36.0,
|
| 603 |
+
"learning_rate": 1.3881621975147155e-05,
|
| 604 |
+
"loss": 28.0072,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.042188659344972006,
|
| 609 |
+
"grad_norm": 49.75,
|
| 610 |
+
"learning_rate": 1.404512753433617e-05,
|
| 611 |
+
"loss": 26.4452,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.04267922515130889,
|
| 616 |
+
"grad_norm": 49.5,
|
| 617 |
+
"learning_rate": 1.420863309352518e-05,
|
| 618 |
+
"loss": 27.5915,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.043169790957645775,
|
| 623 |
+
"grad_norm": 45.5,
|
| 624 |
+
"learning_rate": 1.4372138652714192e-05,
|
| 625 |
+
"loss": 25.8689,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.04366035676398266,
|
| 630 |
+
"grad_norm": 75.0,
|
| 631 |
+
"learning_rate": 1.4535644211903204e-05,
|
| 632 |
+
"loss": 26.5536,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.044150922570319544,
|
| 637 |
+
"grad_norm": 38.0,
|
| 638 |
+
"learning_rate": 1.4699149771092219e-05,
|
| 639 |
+
"loss": 29.0372,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.04464148837665643,
|
| 644 |
+
"grad_norm": 36.25,
|
| 645 |
+
"learning_rate": 1.486265533028123e-05,
|
| 646 |
+
"loss": 25.1781,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.04513205418299331,
|
| 651 |
+
"grad_norm": 34.5,
|
| 652 |
+
"learning_rate": 1.5026160889470242e-05,
|
| 653 |
+
"loss": 25.2917,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.04562261998933019,
|
| 658 |
+
"grad_norm": 30.0,
|
| 659 |
+
"learning_rate": 1.5189666448659254e-05,
|
| 660 |
+
"loss": 24.7534,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.046113185795667075,
|
| 665 |
+
"grad_norm": 20.75,
|
| 666 |
+
"learning_rate": 1.535317200784827e-05,
|
| 667 |
+
"loss": 24.7541,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.04660375160200396,
|
| 672 |
+
"grad_norm": 111.5,
|
| 673 |
+
"learning_rate": 1.551667756703728e-05,
|
| 674 |
+
"loss": 25.0763,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.047094317408340844,
|
| 679 |
+
"grad_norm": 35.75,
|
| 680 |
+
"learning_rate": 1.5680183126226293e-05,
|
| 681 |
+
"loss": 23.9253,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.04758488321467773,
|
| 686 |
+
"grad_norm": 23.375,
|
| 687 |
+
"learning_rate": 1.5843688685415306e-05,
|
| 688 |
+
"loss": 22.4641,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.04807544902101461,
|
| 693 |
+
"grad_norm": 18.125,
|
| 694 |
+
"learning_rate": 1.6007194244604316e-05,
|
| 695 |
+
"loss": 21.8752,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.0485660148273515,
|
| 700 |
+
"grad_norm": 19.375,
|
| 701 |
+
"learning_rate": 1.617069980379333e-05,
|
| 702 |
+
"loss": 20.6978,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.04905658063368838,
|
| 707 |
+
"grad_norm": 15.5625,
|
| 708 |
+
"learning_rate": 1.633420536298234e-05,
|
| 709 |
+
"loss": 19.1208,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.049547146440025266,
|
| 714 |
+
"grad_norm": 31.0,
|
| 715 |
+
"learning_rate": 1.6497710922171357e-05,
|
| 716 |
+
"loss": 19.9687,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.05003771224636215,
|
| 721 |
+
"grad_norm": 94.5,
|
| 722 |
+
"learning_rate": 1.6661216481360367e-05,
|
| 723 |
+
"loss": 19.8846,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.05052827805269903,
|
| 728 |
+
"grad_norm": 167.0,
|
| 729 |
+
"learning_rate": 1.682472204054938e-05,
|
| 730 |
+
"loss": 20.3119,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.05101884385903591,
|
| 735 |
+
"grad_norm": 31.875,
|
| 736 |
+
"learning_rate": 1.6988227599738393e-05,
|
| 737 |
+
"loss": 19.763,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.0515094096653728,
|
| 742 |
+
"grad_norm": 30.125,
|
| 743 |
+
"learning_rate": 1.7151733158927403e-05,
|
| 744 |
+
"loss": 18.4537,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.05199997547170968,
|
| 749 |
+
"grad_norm": 28.75,
|
| 750 |
+
"learning_rate": 1.7315238718116417e-05,
|
| 751 |
+
"loss": 18.8536,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.05249054127804657,
|
| 756 |
+
"grad_norm": 47.5,
|
| 757 |
+
"learning_rate": 1.7478744277305427e-05,
|
| 758 |
+
"loss": 19.0843,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.05298110708438345,
|
| 763 |
+
"grad_norm": 10.875,
|
| 764 |
+
"learning_rate": 1.7642249836494444e-05,
|
| 765 |
+
"loss": 18.1357,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.053471672890720336,
|
| 770 |
+
"grad_norm": 12.375,
|
| 771 |
+
"learning_rate": 1.7805755395683454e-05,
|
| 772 |
+
"loss": 17.805,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.05396223869705722,
|
| 777 |
+
"grad_norm": 23.5,
|
| 778 |
+
"learning_rate": 1.7969260954872467e-05,
|
| 779 |
+
"loss": 17.2343,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.054452804503394105,
|
| 784 |
+
"grad_norm": 16.0,
|
| 785 |
+
"learning_rate": 1.8132766514061477e-05,
|
| 786 |
+
"loss": 19.1025,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.05494337030973099,
|
| 791 |
+
"grad_norm": 13.4375,
|
| 792 |
+
"learning_rate": 1.829627207325049e-05,
|
| 793 |
+
"loss": 17.2674,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.05543393611606787,
|
| 798 |
+
"grad_norm": 34.25,
|
| 799 |
+
"learning_rate": 1.8459777632439504e-05,
|
| 800 |
+
"loss": 17.5344,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.05592450192240475,
|
| 805 |
+
"grad_norm": 23.375,
|
| 806 |
+
"learning_rate": 1.8623283191628514e-05,
|
| 807 |
+
"loss": 16.709,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.056415067728741636,
|
| 812 |
+
"grad_norm": 27.125,
|
| 813 |
+
"learning_rate": 1.878678875081753e-05,
|
| 814 |
+
"loss": 17.5936,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.05690563353507852,
|
| 819 |
+
"grad_norm": 8.6875,
|
| 820 |
+
"learning_rate": 1.895029431000654e-05,
|
| 821 |
+
"loss": 18.3802,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.057396199341415405,
|
| 826 |
+
"grad_norm": 14.5625,
|
| 827 |
+
"learning_rate": 1.9113799869195555e-05,
|
| 828 |
+
"loss": 16.4857,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.05788676514775229,
|
| 833 |
+
"grad_norm": 23.25,
|
| 834 |
+
"learning_rate": 1.9277305428384565e-05,
|
| 835 |
+
"loss": 15.4465,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.058377330954089174,
|
| 840 |
+
"grad_norm": 13.6875,
|
| 841 |
+
"learning_rate": 1.9440810987573578e-05,
|
| 842 |
+
"loss": 15.4513,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.05886789676042606,
|
| 847 |
+
"grad_norm": 45.5,
|
| 848 |
+
"learning_rate": 1.960431654676259e-05,
|
| 849 |
+
"loss": 15.0515,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.05935846256676294,
|
| 854 |
+
"grad_norm": 17.625,
|
| 855 |
+
"learning_rate": 1.97678221059516e-05,
|
| 856 |
+
"loss": 17.6995,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.05984902837309983,
|
| 861 |
+
"grad_norm": 26.375,
|
| 862 |
+
"learning_rate": 1.9931327665140615e-05,
|
| 863 |
+
"loss": 15.2606,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.060339594179436705,
|
| 868 |
+
"grad_norm": 11.0625,
|
| 869 |
+
"learning_rate": 2.009483322432963e-05,
|
| 870 |
+
"loss": 15.909,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.06083015998577359,
|
| 875 |
+
"grad_norm": 260.0,
|
| 876 |
+
"learning_rate": 2.0258338783518642e-05,
|
| 877 |
+
"loss": 14.311,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.061320725792110474,
|
| 882 |
+
"grad_norm": 9.4375,
|
| 883 |
+
"learning_rate": 2.0421844342707652e-05,
|
| 884 |
+
"loss": 13.3078,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.06181129159844736,
|
| 889 |
+
"grad_norm": 113.0,
|
| 890 |
+
"learning_rate": 2.0585349901896665e-05,
|
| 891 |
+
"loss": 14.7056,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.06230185740478424,
|
| 896 |
+
"grad_norm": 10.5625,
|
| 897 |
+
"learning_rate": 2.074885546108568e-05,
|
| 898 |
+
"loss": 13.4457,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.06279242321112112,
|
| 903 |
+
"grad_norm": 10.5625,
|
| 904 |
+
"learning_rate": 2.091236102027469e-05,
|
| 905 |
+
"loss": 13.3094,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.063282989017458,
|
| 910 |
+
"grad_norm": 22.625,
|
| 911 |
+
"learning_rate": 2.1075866579463702e-05,
|
| 912 |
+
"loss": 13.7567,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.06377355482379489,
|
| 917 |
+
"grad_norm": 12.625,
|
| 918 |
+
"learning_rate": 2.1239372138652716e-05,
|
| 919 |
+
"loss": 14.2653,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.06426412063013177,
|
| 924 |
+
"grad_norm": 143.0,
|
| 925 |
+
"learning_rate": 2.140287769784173e-05,
|
| 926 |
+
"loss": 14.4901,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.06475468643646866,
|
| 931 |
+
"grad_norm": 13.1875,
|
| 932 |
+
"learning_rate": 2.156638325703074e-05,
|
| 933 |
+
"loss": 13.5675,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.06524525224280554,
|
| 938 |
+
"grad_norm": 4.34375,
|
| 939 |
+
"learning_rate": 2.1729888816219753e-05,
|
| 940 |
+
"loss": 13.0833,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.06573581804914243,
|
| 945 |
+
"grad_norm": 63.25,
|
| 946 |
+
"learning_rate": 2.1893394375408766e-05,
|
| 947 |
+
"loss": 14.6539,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.06622638385547931,
|
| 952 |
+
"grad_norm": 6.78125,
|
| 953 |
+
"learning_rate": 2.2056899934597776e-05,
|
| 954 |
+
"loss": 12.8027,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.0667169496618162,
|
| 959 |
+
"grad_norm": 318.0,
|
| 960 |
+
"learning_rate": 2.222040549378679e-05,
|
| 961 |
+
"loss": 12.5888,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.06720751546815308,
|
| 966 |
+
"grad_norm": 5.5625,
|
| 967 |
+
"learning_rate": 2.2383911052975803e-05,
|
| 968 |
+
"loss": 12.7075,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.06769808127448997,
|
| 973 |
+
"grad_norm": 57.75,
|
| 974 |
+
"learning_rate": 2.2547416612164817e-05,
|
| 975 |
+
"loss": 13.2399,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.06818864708082685,
|
| 980 |
+
"grad_norm": 6.78125,
|
| 981 |
+
"learning_rate": 2.2710922171353827e-05,
|
| 982 |
+
"loss": 13.3551,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.06867921288716373,
|
| 987 |
+
"grad_norm": 4.78125,
|
| 988 |
+
"learning_rate": 2.287442773054284e-05,
|
| 989 |
+
"loss": 12.7103,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.06916977869350062,
|
| 994 |
+
"grad_norm": 5.09375,
|
| 995 |
+
"learning_rate": 2.3037933289731853e-05,
|
| 996 |
+
"loss": 12.9241,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.0696603444998375,
|
| 1001 |
+
"grad_norm": 10.5625,
|
| 1002 |
+
"learning_rate": 2.3201438848920864e-05,
|
| 1003 |
+
"loss": 12.9225,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.07015091030617439,
|
| 1008 |
+
"grad_norm": 5.8125,
|
| 1009 |
+
"learning_rate": 2.3364944408109877e-05,
|
| 1010 |
+
"loss": 12.943,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.07064147611251127,
|
| 1015 |
+
"grad_norm": 6.75,
|
| 1016 |
+
"learning_rate": 2.352844996729889e-05,
|
| 1017 |
+
"loss": 14.0352,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.07113204191884816,
|
| 1022 |
+
"grad_norm": 5.53125,
|
| 1023 |
+
"learning_rate": 2.3691955526487904e-05,
|
| 1024 |
+
"loss": 12.5482,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.07162260772518504,
|
| 1029 |
+
"grad_norm": 3.625,
|
| 1030 |
+
"learning_rate": 2.3855461085676914e-05,
|
| 1031 |
+
"loss": 11.5144,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.07211317353152191,
|
| 1036 |
+
"grad_norm": 164.0,
|
| 1037 |
+
"learning_rate": 2.4018966644865924e-05,
|
| 1038 |
+
"loss": 12.4646,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.0726037393378588,
|
| 1043 |
+
"grad_norm": 40.0,
|
| 1044 |
+
"learning_rate": 2.418247220405494e-05,
|
| 1045 |
+
"loss": 12.4292,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.07309430514419568,
|
| 1050 |
+
"grad_norm": 19.125,
|
| 1051 |
+
"learning_rate": 2.434597776324395e-05,
|
| 1052 |
+
"loss": 12.3798,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.07358487095053257,
|
| 1057 |
+
"grad_norm": 229.0,
|
| 1058 |
+
"learning_rate": 2.4509483322432964e-05,
|
| 1059 |
+
"loss": 12.9266,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.07407543675686945,
|
| 1064 |
+
"grad_norm": 12.0,
|
| 1065 |
+
"learning_rate": 2.4672988881621974e-05,
|
| 1066 |
+
"loss": 11.541,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.07456600256320634,
|
| 1071 |
+
"grad_norm": 10.125,
|
| 1072 |
+
"learning_rate": 2.483649444081099e-05,
|
| 1073 |
+
"loss": 12.6567,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.07505656836954322,
|
| 1078 |
+
"grad_norm": 4.03125,
|
| 1079 |
+
"learning_rate": 2.5e-05,
|
| 1080 |
+
"loss": 11.67,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.0755471341758801,
|
| 1085 |
+
"grad_norm": 25.625,
|
| 1086 |
+
"learning_rate": 2.5163505559189015e-05,
|
| 1087 |
+
"loss": 11.3946,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.07603769998221699,
|
| 1092 |
+
"grad_norm": 13.5,
|
| 1093 |
+
"learning_rate": 2.5327011118378025e-05,
|
| 1094 |
+
"loss": 11.414,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.07652826578855387,
|
| 1099 |
+
"grad_norm": 4.96875,
|
| 1100 |
+
"learning_rate": 2.5490516677567038e-05,
|
| 1101 |
+
"loss": 11.6602,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.07701883159489076,
|
| 1106 |
+
"grad_norm": 3.09375,
|
| 1107 |
+
"learning_rate": 2.5654022236756055e-05,
|
| 1108 |
+
"loss": 11.4031,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.07750939740122764,
|
| 1113 |
+
"grad_norm": 6.96875,
|
| 1114 |
+
"learning_rate": 2.581752779594506e-05,
|
| 1115 |
+
"loss": 10.9844,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.07799996320756453,
|
| 1120 |
+
"grad_norm": 3.28125,
|
| 1121 |
+
"learning_rate": 2.598103335513408e-05,
|
| 1122 |
+
"loss": 12.9034,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.07849052901390141,
|
| 1127 |
+
"grad_norm": 5.59375,
|
| 1128 |
+
"learning_rate": 2.6144538914323085e-05,
|
| 1129 |
+
"loss": 11.7687,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.0789810948202383,
|
| 1134 |
+
"grad_norm": 133.0,
|
| 1135 |
+
"learning_rate": 2.63080444735121e-05,
|
| 1136 |
+
"loss": 12.0474,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.07947166062657518,
|
| 1141 |
+
"grad_norm": 8.1875,
|
| 1142 |
+
"learning_rate": 2.6471550032701115e-05,
|
| 1143 |
+
"loss": 11.8777,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.07996222643291206,
|
| 1148 |
+
"grad_norm": 3.78125,
|
| 1149 |
+
"learning_rate": 2.6635055591890122e-05,
|
| 1150 |
+
"loss": 11.8737,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.08045279223924895,
|
| 1155 |
+
"grad_norm": 3.53125,
|
| 1156 |
+
"learning_rate": 2.679856115107914e-05,
|
| 1157 |
+
"loss": 12.319,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.08094335804558583,
|
| 1162 |
+
"grad_norm": 4.59375,
|
| 1163 |
+
"learning_rate": 2.6962066710268152e-05,
|
| 1164 |
+
"loss": 11.686,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.08143392385192272,
|
| 1169 |
+
"grad_norm": 2.765625,
|
| 1170 |
+
"learning_rate": 2.7125572269457162e-05,
|
| 1171 |
+
"loss": 10.6974,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.08192448965825959,
|
| 1176 |
+
"grad_norm": 4.40625,
|
| 1177 |
+
"learning_rate": 2.7289077828646176e-05,
|
| 1178 |
+
"loss": 12.4035,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.08241505546459647,
|
| 1183 |
+
"grad_norm": 3.109375,
|
| 1184 |
+
"learning_rate": 2.745258338783519e-05,
|
| 1185 |
+
"loss": 11.7501,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.08290562127093336,
|
| 1190 |
+
"grad_norm": 5.34375,
|
| 1191 |
+
"learning_rate": 2.76160889470242e-05,
|
| 1192 |
+
"loss": 11.4395,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.08339618707727024,
|
| 1197 |
+
"grad_norm": 12.8125,
|
| 1198 |
+
"learning_rate": 2.7779594506213213e-05,
|
| 1199 |
+
"loss": 11.7575,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.08388675288360713,
|
| 1204 |
+
"grad_norm": 7.1875,
|
| 1205 |
+
"learning_rate": 2.7943100065402223e-05,
|
| 1206 |
+
"loss": 11.2958,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.08437731868994401,
|
| 1211 |
+
"grad_norm": 2.328125,
|
| 1212 |
+
"learning_rate": 2.8106605624591236e-05,
|
| 1213 |
+
"loss": 11.4801,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.0848678844962809,
|
| 1218 |
+
"grad_norm": 6.5625,
|
| 1219 |
+
"learning_rate": 2.827011118378025e-05,
|
| 1220 |
+
"loss": 11.5054,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.08535845030261778,
|
| 1225 |
+
"grad_norm": 2.546875,
|
| 1226 |
+
"learning_rate": 2.843361674296926e-05,
|
| 1227 |
+
"loss": 11.1001,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 0.08584901610895466,
|
| 1232 |
+
"grad_norm": 2.359375,
|
| 1233 |
+
"learning_rate": 2.8597122302158273e-05,
|
| 1234 |
+
"loss": 12.427,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 0.08633958191529155,
|
| 1239 |
+
"grad_norm": 7.40625,
|
| 1240 |
+
"learning_rate": 2.876062786134729e-05,
|
| 1241 |
+
"loss": 11.2234,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.08683014772162843,
|
| 1246 |
+
"grad_norm": 248.0,
|
| 1247 |
+
"learning_rate": 2.8924133420536297e-05,
|
| 1248 |
+
"loss": 11.7346,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.08732071352796532,
|
| 1253 |
+
"grad_norm": 10.125,
|
| 1254 |
+
"learning_rate": 2.9087638979725313e-05,
|
| 1255 |
+
"loss": 12.5661,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.0878112793343022,
|
| 1260 |
+
"grad_norm": 2.484375,
|
| 1261 |
+
"learning_rate": 2.9251144538914327e-05,
|
| 1262 |
+
"loss": 10.1607,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.08830184514063909,
|
| 1267 |
+
"grad_norm": 2.65625,
|
| 1268 |
+
"learning_rate": 2.9414650098103337e-05,
|
| 1269 |
+
"loss": 11.1169,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.08879241094697597,
|
| 1274 |
+
"grad_norm": 2.625,
|
| 1275 |
+
"learning_rate": 2.957815565729235e-05,
|
| 1276 |
+
"loss": 10.876,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 0.08928297675331286,
|
| 1281 |
+
"grad_norm": 3.046875,
|
| 1282 |
+
"learning_rate": 2.974166121648136e-05,
|
| 1283 |
+
"loss": 11.7584,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 0.08977354255964974,
|
| 1288 |
+
"grad_norm": 2.296875,
|
| 1289 |
+
"learning_rate": 2.9905166775670374e-05,
|
| 1290 |
+
"loss": 11.4279,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.09026410836598663,
|
| 1295 |
+
"grad_norm": 4.125,
|
| 1296 |
+
"learning_rate": 3.0068672334859387e-05,
|
| 1297 |
+
"loss": 10.9975,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.09075467417232351,
|
| 1302 |
+
"grad_norm": 3.234375,
|
| 1303 |
+
"learning_rate": 3.0232177894048397e-05,
|
| 1304 |
+
"loss": 10.9377,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.09124523997866038,
|
| 1309 |
+
"grad_norm": 4.34375,
|
| 1310 |
+
"learning_rate": 3.039568345323741e-05,
|
| 1311 |
+
"loss": 11.1918,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 0.09173580578499727,
|
| 1316 |
+
"grad_norm": 2.59375,
|
| 1317 |
+
"learning_rate": 3.055918901242643e-05,
|
| 1318 |
+
"loss": 10.8162,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.09222637159133415,
|
| 1323 |
+
"grad_norm": 3.84375,
|
| 1324 |
+
"learning_rate": 3.0722694571615434e-05,
|
| 1325 |
+
"loss": 11.1515,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 0.09271693739767103,
|
| 1330 |
+
"grad_norm": 5.46875,
|
| 1331 |
+
"learning_rate": 3.088620013080445e-05,
|
| 1332 |
+
"loss": 11.5057,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 0.09320750320400792,
|
| 1337 |
+
"grad_norm": 3.328125,
|
| 1338 |
+
"learning_rate": 3.104970568999346e-05,
|
| 1339 |
+
"loss": 11.1651,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.0936980690103448,
|
| 1344 |
+
"grad_norm": 2.40625,
|
| 1345 |
+
"learning_rate": 3.1213211249182475e-05,
|
| 1346 |
+
"loss": 11.0284,
|
| 1347 |
+
"step": 1910
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.09418863481668169,
|
| 1351 |
+
"grad_norm": 3.765625,
|
| 1352 |
+
"learning_rate": 3.137671680837149e-05,
|
| 1353 |
+
"loss": 11.2881,
|
| 1354 |
+
"step": 1920
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.09467920062301857,
|
| 1358 |
+
"grad_norm": 3.03125,
|
| 1359 |
+
"learning_rate": 3.15402223675605e-05,
|
| 1360 |
+
"loss": 11.6501,
|
| 1361 |
+
"step": 1930
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.09516976642935546,
|
| 1365 |
+
"grad_norm": 2.5,
|
| 1366 |
+
"learning_rate": 3.170372792674951e-05,
|
| 1367 |
+
"loss": 11.2798,
|
| 1368 |
+
"step": 1940
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 0.09566033223569234,
|
| 1372 |
+
"grad_norm": 3.390625,
|
| 1373 |
+
"learning_rate": 3.186723348593852e-05,
|
| 1374 |
+
"loss": 11.6165,
|
| 1375 |
+
"step": 1950
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.09615089804202923,
|
| 1379 |
+
"grad_norm": 1.2734375,
|
| 1380 |
+
"learning_rate": 3.2030739045127535e-05,
|
| 1381 |
+
"loss": 10.3709,
|
| 1382 |
+
"step": 1960
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.09664146384836611,
|
| 1386 |
+
"grad_norm": 2.1875,
|
| 1387 |
+
"learning_rate": 3.219424460431655e-05,
|
| 1388 |
+
"loss": 10.8632,
|
| 1389 |
+
"step": 1970
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.097132029654703,
|
| 1393 |
+
"grad_norm": 7.0625,
|
| 1394 |
+
"learning_rate": 3.235775016350556e-05,
|
| 1395 |
+
"loss": 11.4351,
|
| 1396 |
+
"step": 1980
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 0.09762259546103988,
|
| 1400 |
+
"grad_norm": 1.46875,
|
| 1401 |
+
"learning_rate": 3.252125572269457e-05,
|
| 1402 |
+
"loss": 10.9888,
|
| 1403 |
+
"step": 1990
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 0.09811316126737676,
|
| 1407 |
+
"grad_norm": 1.578125,
|
| 1408 |
+
"learning_rate": 3.268476128188359e-05,
|
| 1409 |
+
"loss": 9.9517,
|
| 1410 |
+
"step": 2000
|
| 1411 |
+
}
|
| 1412 |
+
],
|
| 1413 |
+
"logging_steps": 10,
|
| 1414 |
+
"max_steps": 61155,
|
| 1415 |
+
"num_input_tokens_seen": 0,
|
| 1416 |
+
"num_train_epochs": 3,
|
| 1417 |
+
"save_steps": 2000,
|
| 1418 |
+
"stateful_callbacks": {
|
| 1419 |
+
"TrainerControl": {
|
| 1420 |
+
"args": {
|
| 1421 |
+
"should_epoch_stop": false,
|
| 1422 |
+
"should_evaluate": false,
|
| 1423 |
+
"should_log": false,
|
| 1424 |
+
"should_save": true,
|
| 1425 |
+
"should_training_stop": false
|
| 1426 |
+
},
|
| 1427 |
+
"attributes": {}
|
| 1428 |
+
}
|
| 1429 |
+
},
|
| 1430 |
+
"total_flos": 5.38032432218112e+18,
|
| 1431 |
+
"train_batch_size": 2,
|
| 1432 |
+
"trial_name": null,
|
| 1433 |
+
"trial_params": null
|
| 1434 |
+
}
|
checkpoints/checkpoint-2000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3426ffd33c60179729f2d6634f33740dda1785184d4955f3ba37cecf1acc3a35
|
| 3 |
+
size 5304
|
logs/events.out.tfevents.1752345702.5f638d684589.1618.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:83ef83385d54d4e925289e1586f37cf245b45b119aa165c398e47d02844e012f
|
| 3 |
+
size 47447
|
myolmoe/__pycache__/__init__.cpython-311.pyc
CHANGED
|
Binary files a/myolmoe/__pycache__/__init__.cpython-311.pyc and b/myolmoe/__pycache__/__init__.cpython-311.pyc differ
|
|
|
myolmoe/__pycache__/modeling_myolmoe.cpython-311.pyc
CHANGED
|
Binary files a/myolmoe/__pycache__/modeling_myolmoe.cpython-311.pyc and b/myolmoe/__pycache__/modeling_myolmoe.cpython-311.pyc differ
|
|
|
scripts/__pycache__/train.cpython-311.pyc
CHANGED
|
Binary files a/scripts/__pycache__/train.cpython-311.pyc and b/scripts/__pycache__/train.cpython-311.pyc differ
|
|
|