SykoLLM v6.7
Browse files- config.json +34 -0
- generation_config.json +10 -0
- model.safetensors +3 -0
- tokenizer.json +0 -0
- tokenizer_config.json +11 -0
- trainer_state.json +2064 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"Phi3ForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_dropout": 0.0,
|
| 6 |
+
"bos_token_id": 0,
|
| 7 |
+
"dtype": "float32",
|
| 8 |
+
"embd_pdrop": 0.0,
|
| 9 |
+
"eos_token_id": 0,
|
| 10 |
+
"hidden_act": "silu",
|
| 11 |
+
"hidden_size": 1024,
|
| 12 |
+
"initializer_range": 0.02,
|
| 13 |
+
"intermediate_size": 3072,
|
| 14 |
+
"max_position_embeddings": 1024,
|
| 15 |
+
"model_type": "phi3",
|
| 16 |
+
"num_attention_heads": 8,
|
| 17 |
+
"num_hidden_layers": 24,
|
| 18 |
+
"num_key_value_heads": 2,
|
| 19 |
+
"original_max_position_embeddings": 4096,
|
| 20 |
+
"pad_token_id": 4,
|
| 21 |
+
"partial_rotary_factor": 1.0,
|
| 22 |
+
"resid_pdrop": 0.0,
|
| 23 |
+
"rms_norm_eps": 1e-05,
|
| 24 |
+
"rope_parameters": {
|
| 25 |
+
"partial_rotary_factor": 1.0,
|
| 26 |
+
"rope_theta": 10000.0,
|
| 27 |
+
"rope_type": "default"
|
| 28 |
+
},
|
| 29 |
+
"sliding_window": null,
|
| 30 |
+
"tie_word_embeddings": false,
|
| 31 |
+
"transformers_version": "5.9.0",
|
| 32 |
+
"use_cache": false,
|
| 33 |
+
"vocab_size": 50000
|
| 34 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 0,
|
| 4 |
+
"eos_token_id": 0,
|
| 5 |
+
"output_attentions": false,
|
| 6 |
+
"output_hidden_states": false,
|
| 7 |
+
"pad_token_id": 4,
|
| 8 |
+
"transformers_version": "5.9.0",
|
| 9 |
+
"use_cache": false
|
| 10 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dac0f22a2af395c6c3cb3581fd2d55cdcdb3ca10e62ca3aead097cac5a447d67
|
| 3 |
+
size 1567445472
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"backend": "tokenizers",
|
| 3 |
+
"bos_token": "<|endoftext|>",
|
| 4 |
+
"eos_token": "<|endoftext|>",
|
| 5 |
+
"is_local": false,
|
| 6 |
+
"local_files_only": false,
|
| 7 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 8 |
+
"pad_token": "<|endoftext|>",
|
| 9 |
+
"tokenizer_class": "TokenizersBackend",
|
| 10 |
+
"unk_token": "<|endoftext|>"
|
| 11 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2064 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.118375,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 2900,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.00125,
|
| 14 |
+
"grad_norm": 0.36102330684661865,
|
| 15 |
+
"learning_rate": 5.5665e-06,
|
| 16 |
+
"loss": 2.681707572937012,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0025,
|
| 21 |
+
"grad_norm": 0.34577861428260803,
|
| 22 |
+
"learning_rate": 1.17515e-05,
|
| 23 |
+
"loss": 2.6720260620117187,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.00375,
|
| 28 |
+
"grad_norm": 0.3295978009700775,
|
| 29 |
+
"learning_rate": 1.79365e-05,
|
| 30 |
+
"loss": 2.672147750854492,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.005,
|
| 35 |
+
"grad_norm": 0.32688695192337036,
|
| 36 |
+
"learning_rate": 2.41215e-05,
|
| 37 |
+
"loss": 2.675041389465332,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.00625,
|
| 42 |
+
"grad_norm": 0.3257655203342438,
|
| 43 |
+
"learning_rate": 3.03065e-05,
|
| 44 |
+
"loss": 2.675174522399902,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.0075,
|
| 49 |
+
"grad_norm": 0.336309552192688,
|
| 50 |
+
"learning_rate": 3.6491499999999994e-05,
|
| 51 |
+
"loss": 2.6966915130615234,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.00875,
|
| 56 |
+
"grad_norm": 0.3346744179725647,
|
| 57 |
+
"learning_rate": 4.26765e-05,
|
| 58 |
+
"loss": 2.6632720947265627,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.01,
|
| 63 |
+
"grad_norm": 0.32752859592437744,
|
| 64 |
+
"learning_rate": 4.88615e-05,
|
| 65 |
+
"loss": 2.695608139038086,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.01125,
|
| 70 |
+
"grad_norm": 0.32975664734840393,
|
| 71 |
+
"learning_rate": 5.50465e-05,
|
| 72 |
+
"loss": 2.6731294631958007,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.0125,
|
| 77 |
+
"grad_norm": 0.33192330598831177,
|
| 78 |
+
"learning_rate": 6.12315e-05,
|
| 79 |
+
"loss": 2.6482282638549806,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.01375,
|
| 84 |
+
"grad_norm": 0.3244248032569885,
|
| 85 |
+
"learning_rate": 6.74165e-05,
|
| 86 |
+
"loss": 2.700460433959961,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.015,
|
| 91 |
+
"grad_norm": 0.32452520728111267,
|
| 92 |
+
"learning_rate": 7.36015e-05,
|
| 93 |
+
"loss": 2.6673652648925783,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.01625,
|
| 98 |
+
"grad_norm": 0.32950156927108765,
|
| 99 |
+
"learning_rate": 7.97865e-05,
|
| 100 |
+
"loss": 2.66015510559082,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.0175,
|
| 105 |
+
"grad_norm": 0.3157300651073456,
|
| 106 |
+
"learning_rate": 8.597149999999999e-05,
|
| 107 |
+
"loss": 2.653401184082031,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.01875,
|
| 112 |
+
"grad_norm": 0.3447306156158447,
|
| 113 |
+
"learning_rate": 9.21565e-05,
|
| 114 |
+
"loss": 2.638433837890625,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.02,
|
| 119 |
+
"grad_norm": 0.33060336112976074,
|
| 120 |
+
"learning_rate": 9.834150000000001e-05,
|
| 121 |
+
"loss": 2.6599313735961916,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.02125,
|
| 126 |
+
"grad_norm": 0.33370116353034973,
|
| 127 |
+
"learning_rate": 0.00010452649999999999,
|
| 128 |
+
"loss": 2.675436019897461,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.0225,
|
| 133 |
+
"grad_norm": 0.32309311628341675,
|
| 134 |
+
"learning_rate": 0.0001107115,
|
| 135 |
+
"loss": 2.682134246826172,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.02375,
|
| 140 |
+
"grad_norm": 0.3298942446708679,
|
| 141 |
+
"learning_rate": 0.0001168965,
|
| 142 |
+
"loss": 2.6672037124633787,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.025,
|
| 147 |
+
"grad_norm": 0.3257051408290863,
|
| 148 |
+
"learning_rate": 0.0001230815,
|
| 149 |
+
"loss": 2.6710464477539064,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.02625,
|
| 154 |
+
"grad_norm": 0.32734546065330505,
|
| 155 |
+
"learning_rate": 0.00012369959364576377,
|
| 156 |
+
"loss": 2.6694522857666017,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.0275,
|
| 161 |
+
"grad_norm": 0.3286871016025543,
|
| 162 |
+
"learning_rate": 0.00012369818897130838,
|
| 163 |
+
"loss": 2.67569580078125,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.02875,
|
| 168 |
+
"grad_norm": 0.3206029534339905,
|
| 169 |
+
"learning_rate": 0.0001236957809826964,
|
| 170 |
+
"loss": 2.671968460083008,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.03,
|
| 175 |
+
"grad_norm": 0.32244956493377686,
|
| 176 |
+
"learning_rate": 0.0001236923697189907,
|
| 177 |
+
"loss": 2.6653528213500977,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.03125,
|
| 182 |
+
"grad_norm": 0.3286353647708893,
|
| 183 |
+
"learning_rate": 0.00012368795523552952,
|
| 184 |
+
"loss": 2.644626998901367,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.0325,
|
| 189 |
+
"grad_norm": 0.31846532225608826,
|
| 190 |
+
"learning_rate": 0.00012368253760392556,
|
| 191 |
+
"loss": 2.6661434173583984,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.03375,
|
| 196 |
+
"grad_norm": 0.34063664078712463,
|
| 197 |
+
"learning_rate": 0.00012367611691206466,
|
| 198 |
+
"loss": 2.658544921875,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.035,
|
| 203 |
+
"grad_norm": 0.3394038677215576,
|
| 204 |
+
"learning_rate": 0.00012366869326410474,
|
| 205 |
+
"loss": 2.671076202392578,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.03625,
|
| 210 |
+
"grad_norm": 0.3454046845436096,
|
| 211 |
+
"learning_rate": 0.00012366026678047368,
|
| 212 |
+
"loss": 2.690570068359375,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.0375,
|
| 217 |
+
"grad_norm": 0.32945406436920166,
|
| 218 |
+
"learning_rate": 0.00012365083759786766,
|
| 219 |
+
"loss": 2.6626564025878907,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.03875,
|
| 224 |
+
"grad_norm": 0.3266613483428955,
|
| 225 |
+
"learning_rate": 0.00012364040586924886,
|
| 226 |
+
"loss": 2.6811601638793947,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.04,
|
| 231 |
+
"grad_norm": 0.32137027382850647,
|
| 232 |
+
"learning_rate": 0.0001236289717638429,
|
| 233 |
+
"loss": 2.656772422790527,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.04125,
|
| 238 |
+
"grad_norm": 0.31430286169052124,
|
| 239 |
+
"learning_rate": 0.00012361653546713627,
|
| 240 |
+
"loss": 2.667566680908203,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.0425,
|
| 245 |
+
"grad_norm": 0.3187640905380249,
|
| 246 |
+
"learning_rate": 0.00012360309718087312,
|
| 247 |
+
"loss": 2.6774127960205076,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.04375,
|
| 252 |
+
"grad_norm": 0.3238705098628998,
|
| 253 |
+
"learning_rate": 0.00012358865712305212,
|
| 254 |
+
"loss": 2.650909423828125,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.045,
|
| 259 |
+
"grad_norm": 0.3178948163986206,
|
| 260 |
+
"learning_rate": 0.00012357321552792288,
|
| 261 |
+
"loss": 2.6466007232666016,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.04625,
|
| 266 |
+
"grad_norm": 0.3393631875514984,
|
| 267 |
+
"learning_rate": 0.0001235567726459822,
|
| 268 |
+
"loss": 2.6694786071777346,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.0475,
|
| 273 |
+
"grad_norm": 0.33097463846206665,
|
| 274 |
+
"learning_rate": 0.00012353932874396988,
|
| 275 |
+
"loss": 2.6705909729003907,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.04875,
|
| 280 |
+
"grad_norm": 0.3237457275390625,
|
| 281 |
+
"learning_rate": 0.00012352088410486452,
|
| 282 |
+
"loss": 2.666813087463379,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.05,
|
| 287 |
+
"grad_norm": 0.32804396748542786,
|
| 288 |
+
"learning_rate": 0.0001235014390278789,
|
| 289 |
+
"loss": 2.6341262817382813,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.05125,
|
| 294 |
+
"grad_norm": 0.3098997473716736,
|
| 295 |
+
"learning_rate": 0.0001234809938284551,
|
| 296 |
+
"loss": 2.653286361694336,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.0525,
|
| 301 |
+
"grad_norm": 0.31869447231292725,
|
| 302 |
+
"learning_rate": 0.00012345954883825937,
|
| 303 |
+
"loss": 2.6676279067993165,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.05375,
|
| 308 |
+
"grad_norm": 0.3462599813938141,
|
| 309 |
+
"learning_rate": 0.0001234371044051768,
|
| 310 |
+
"loss": 2.6937137603759767,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.055,
|
| 315 |
+
"grad_norm": 0.33410680294036865,
|
| 316 |
+
"learning_rate": 0.00012341366089330566,
|
| 317 |
+
"loss": 2.6624752044677735,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.05625,
|
| 322 |
+
"grad_norm": 0.3401891589164734,
|
| 323 |
+
"learning_rate": 0.00012338921868295142,
|
| 324 |
+
"loss": 2.6673324584960936,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.0575,
|
| 329 |
+
"grad_norm": 0.3144513964653015,
|
| 330 |
+
"learning_rate": 0.00012336377817062075,
|
| 331 |
+
"loss": 2.6684280395507813,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.05875,
|
| 336 |
+
"grad_norm": 0.31319352984428406,
|
| 337 |
+
"learning_rate": 0.00012333733976901485,
|
| 338 |
+
"loss": 2.6631874084472655,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.06,
|
| 343 |
+
"grad_norm": 0.3231050670146942,
|
| 344 |
+
"learning_rate": 0.00012330990390702298,
|
| 345 |
+
"loss": 2.6671581268310547,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.06125,
|
| 350 |
+
"grad_norm": 0.3283950686454773,
|
| 351 |
+
"learning_rate": 0.00012328147102971544,
|
| 352 |
+
"loss": 2.6682722091674806,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.0625,
|
| 357 |
+
"grad_norm": 0.3203584849834442,
|
| 358 |
+
"learning_rate": 0.0001232520415983362,
|
| 359 |
+
"loss": 2.6619497299194337,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.06375,
|
| 364 |
+
"grad_norm": 0.3314996063709259,
|
| 365 |
+
"learning_rate": 0.00012322161609029563,
|
| 366 |
+
"loss": 2.675333023071289,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.065,
|
| 371 |
+
"grad_norm": 0.3124040961265564,
|
| 372 |
+
"learning_rate": 0.00012319019499916267,
|
| 373 |
+
"loss": 2.674266052246094,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.06625,
|
| 378 |
+
"grad_norm": 0.334187775850296,
|
| 379 |
+
"learning_rate": 0.0001231577788346567,
|
| 380 |
+
"loss": 2.6644060134887697,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.0675,
|
| 385 |
+
"grad_norm": 0.33853819966316223,
|
| 386 |
+
"learning_rate": 0.00012312436812263953,
|
| 387 |
+
"loss": 2.6285802841186525,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.06875,
|
| 392 |
+
"grad_norm": 0.3226993680000305,
|
| 393 |
+
"learning_rate": 0.00012308996340510664,
|
| 394 |
+
"loss": 2.6620355606079102,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.07,
|
| 399 |
+
"grad_norm": 0.31965890526771545,
|
| 400 |
+
"learning_rate": 0.0001230545652401785,
|
| 401 |
+
"loss": 2.669430160522461,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.07125,
|
| 406 |
+
"grad_norm": 0.323632150888443,
|
| 407 |
+
"learning_rate": 0.00012301817420209152,
|
| 408 |
+
"loss": 2.6710559844970705,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.0725,
|
| 413 |
+
"grad_norm": 0.3202168047428131,
|
| 414 |
+
"learning_rate": 0.00012298079088118863,
|
| 415 |
+
"loss": 2.6743343353271483,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.07375,
|
| 420 |
+
"grad_norm": 0.3278695344924927,
|
| 421 |
+
"learning_rate": 0.00012294241588390982,
|
| 422 |
+
"loss": 2.643411636352539,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.075,
|
| 427 |
+
"grad_norm": 0.3302673101425171,
|
| 428 |
+
"learning_rate": 0.0001229030498327823,
|
| 429 |
+
"loss": 2.7156848907470703,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.07625,
|
| 434 |
+
"grad_norm": 0.31964629888534546,
|
| 435 |
+
"learning_rate": 0.00012286269336641027,
|
| 436 |
+
"loss": 2.6369789123535154,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.0775,
|
| 441 |
+
"grad_norm": 0.32528844475746155,
|
| 442 |
+
"learning_rate": 0.00012282134713946472,
|
| 443 |
+
"loss": 2.655129241943359,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.07875,
|
| 448 |
+
"grad_norm": 0.3346538245677948,
|
| 449 |
+
"learning_rate": 0.00012277901182267275,
|
| 450 |
+
"loss": 2.6634849548339843,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.08,
|
| 455 |
+
"grad_norm": 0.32035592198371887,
|
| 456 |
+
"learning_rate": 0.00012273568810280665,
|
| 457 |
+
"loss": 2.6622406005859376,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.08125,
|
| 462 |
+
"grad_norm": 0.32753705978393555,
|
| 463 |
+
"learning_rate": 0.00012269137668267276,
|
| 464 |
+
"loss": 2.6673862457275392,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.0825,
|
| 469 |
+
"grad_norm": 0.3323623538017273,
|
| 470 |
+
"learning_rate": 0.00012264607828110018,
|
| 471 |
+
"loss": 2.6660182952880858,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.08375,
|
| 476 |
+
"grad_norm": 0.3228432238101959,
|
| 477 |
+
"learning_rate": 0.0001225997936329289,
|
| 478 |
+
"loss": 2.690377044677734,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.085,
|
| 483 |
+
"grad_norm": 0.3340938687324524,
|
| 484 |
+
"learning_rate": 0.00012255252348899816,
|
| 485 |
+
"loss": 2.6579252243041993,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.08625,
|
| 490 |
+
"grad_norm": 0.32717493176460266,
|
| 491 |
+
"learning_rate": 0.00012250426861613406,
|
| 492 |
+
"loss": 2.6669349670410156,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.0875,
|
| 497 |
+
"grad_norm": 0.3213510513305664,
|
| 498 |
+
"learning_rate": 0.0001224550297971371,
|
| 499 |
+
"loss": 2.658818817138672,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.08875,
|
| 504 |
+
"grad_norm": 0.3103785216808319,
|
| 505 |
+
"learning_rate": 0.00012240480783076967,
|
| 506 |
+
"loss": 2.64670467376709,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.09,
|
| 511 |
+
"grad_norm": 0.3206445276737213,
|
| 512 |
+
"learning_rate": 0.00012235360353174288,
|
| 513 |
+
"loss": 2.649314117431641,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.09125,
|
| 518 |
+
"grad_norm": 0.3210267722606659,
|
| 519 |
+
"learning_rate": 0.00012230141773070355,
|
| 520 |
+
"loss": 2.6637636184692384,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.0925,
|
| 525 |
+
"grad_norm": 0.312549352645874,
|
| 526 |
+
"learning_rate": 0.00012224825127422055,
|
| 527 |
+
"loss": 2.6725765228271485,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.09375,
|
| 532 |
+
"grad_norm": 0.32557615637779236,
|
| 533 |
+
"learning_rate": 0.00012219410502477114,
|
| 534 |
+
"loss": 2.6337608337402343,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.095,
|
| 539 |
+
"grad_norm": 0.31713125109672546,
|
| 540 |
+
"learning_rate": 0.00012213897986072705,
|
| 541 |
+
"loss": 2.6361785888671876,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.09625,
|
| 546 |
+
"grad_norm": 0.3173486590385437,
|
| 547 |
+
"learning_rate": 0.00012208287667634017,
|
| 548 |
+
"loss": 2.6491493225097655,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.0975,
|
| 553 |
+
"grad_norm": 0.32202011346817017,
|
| 554 |
+
"learning_rate": 0.00012202579638172791,
|
| 555 |
+
"loss": 2.665495681762695,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.09875,
|
| 560 |
+
"grad_norm": 0.31751731038093567,
|
| 561 |
+
"learning_rate": 0.0001219677399028587,
|
| 562 |
+
"loss": 2.670880889892578,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.1,
|
| 567 |
+
"grad_norm": 0.3310263156890869,
|
| 568 |
+
"learning_rate": 0.00012190870818153682,
|
| 569 |
+
"loss": 2.6745986938476562,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.10125,
|
| 574 |
+
"grad_norm": 0.3246520757675171,
|
| 575 |
+
"learning_rate": 0.00012184870217538704,
|
| 576 |
+
"loss": 2.6367824554443358,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.1025,
|
| 581 |
+
"grad_norm": 0.31728002429008484,
|
| 582 |
+
"learning_rate": 0.0001217877228578393,
|
| 583 |
+
"loss": 2.657224655151367,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.10375,
|
| 588 |
+
"grad_norm": 0.32666370272636414,
|
| 589 |
+
"learning_rate": 0.00012172577121811272,
|
| 590 |
+
"loss": 2.629240798950195,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.105,
|
| 595 |
+
"grad_norm": 0.32864195108413696,
|
| 596 |
+
"learning_rate": 0.00012166284826119965,
|
| 597 |
+
"loss": 2.6314460754394533,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.10625,
|
| 602 |
+
"grad_norm": 0.331391304731369,
|
| 603 |
+
"learning_rate": 0.00012159895500784936,
|
| 604 |
+
"loss": 2.6207229614257814,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.1075,
|
| 609 |
+
"grad_norm": 0.32856595516204834,
|
| 610 |
+
"learning_rate": 0.00012153409249455148,
|
| 611 |
+
"loss": 2.6828586578369142,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.10875,
|
| 616 |
+
"grad_norm": 0.3259557783603668,
|
| 617 |
+
"learning_rate": 0.00012146826177351913,
|
| 618 |
+
"loss": 2.6800840377807615,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.11,
|
| 623 |
+
"grad_norm": 0.3368566930294037,
|
| 624 |
+
"learning_rate": 0.00012140146391267196,
|
| 625 |
+
"loss": 2.644548797607422,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.11125,
|
| 630 |
+
"grad_norm": 0.3319634199142456,
|
| 631 |
+
"learning_rate": 0.00012133369999561872,
|
| 632 |
+
"loss": 2.6457305908203126,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.1125,
|
| 637 |
+
"grad_norm": 0.31302639842033386,
|
| 638 |
+
"learning_rate": 0.00012126497112163972,
|
| 639 |
+
"loss": 2.6418832778930663,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.11375,
|
| 644 |
+
"grad_norm": 0.32079464197158813,
|
| 645 |
+
"learning_rate": 0.00012119527840566905,
|
| 646 |
+
"loss": 2.6311697006225585,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.115,
|
| 651 |
+
"grad_norm": 0.32719048857688904,
|
| 652 |
+
"learning_rate": 0.00012112462297827639,
|
| 653 |
+
"loss": 2.641567611694336,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.11625,
|
| 658 |
+
"grad_norm": 0.32264548540115356,
|
| 659 |
+
"learning_rate": 0.00012105300598564874,
|
| 660 |
+
"loss": 2.6696403503417967,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.1175,
|
| 665 |
+
"grad_norm": 0.3197903335094452,
|
| 666 |
+
"learning_rate": 0.00012098042858957183,
|
| 667 |
+
"loss": 2.6566593170166017,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.11875,
|
| 672 |
+
"grad_norm": 0.3231068253517151,
|
| 673 |
+
"learning_rate": 0.00012090689196741124,
|
| 674 |
+
"loss": 2.63052978515625,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.12,
|
| 679 |
+
"grad_norm": 0.3268223702907562,
|
| 680 |
+
"learning_rate": 0.00012083239731209331,
|
| 681 |
+
"loss": 2.6513845443725588,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.12125,
|
| 686 |
+
"grad_norm": 0.3304605484008789,
|
| 687 |
+
"learning_rate": 0.00012075694583208578,
|
| 688 |
+
"loss": 2.6264434814453126,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.1225,
|
| 693 |
+
"grad_norm": 0.3171931505203247,
|
| 694 |
+
"learning_rate": 0.00012068053875137824,
|
| 695 |
+
"loss": 2.636788558959961,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.12375,
|
| 700 |
+
"grad_norm": 0.3341807425022125,
|
| 701 |
+
"learning_rate": 0.00012060317730946224,
|
| 702 |
+
"loss": 2.6531208038330076,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.125,
|
| 707 |
+
"grad_norm": 0.3334127962589264,
|
| 708 |
+
"learning_rate": 0.00012052486276131108,
|
| 709 |
+
"loss": 2.6705049514770507,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.12625,
|
| 714 |
+
"grad_norm": 0.307980477809906,
|
| 715 |
+
"learning_rate": 0.00012044559637735965,
|
| 716 |
+
"loss": 2.6561138153076174,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.1275,
|
| 721 |
+
"grad_norm": 0.31699395179748535,
|
| 722 |
+
"learning_rate": 0.00012036537944348368,
|
| 723 |
+
"loss": 2.633596420288086,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.12875,
|
| 728 |
+
"grad_norm": 0.32349589467048645,
|
| 729 |
+
"learning_rate": 0.0001202842132609789,
|
| 730 |
+
"loss": 2.651826858520508,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.13,
|
| 735 |
+
"grad_norm": 0.3407875895500183,
|
| 736 |
+
"learning_rate": 0.00012020209914653999,
|
| 737 |
+
"loss": 2.6381755828857423,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.13125,
|
| 742 |
+
"grad_norm": 0.31691980361938477,
|
| 743 |
+
"learning_rate": 0.00012011903843223914,
|
| 744 |
+
"loss": 2.6360122680664064,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.1325,
|
| 749 |
+
"grad_norm": 0.31067660450935364,
|
| 750 |
+
"learning_rate": 0.0001200350324655045,
|
| 751 |
+
"loss": 2.6421882629394533,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.13375,
|
| 756 |
+
"grad_norm": 0.32634156942367554,
|
| 757 |
+
"learning_rate": 0.0001199500826090983,
|
| 758 |
+
"loss": 2.63830509185791,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.135,
|
| 763 |
+
"grad_norm": 0.3369225263595581,
|
| 764 |
+
"learning_rate": 0.00011986419024109472,
|
| 765 |
+
"loss": 2.63408203125,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.13625,
|
| 770 |
+
"grad_norm": 0.3302381932735443,
|
| 771 |
+
"learning_rate": 0.0001197773567548576,
|
| 772 |
+
"loss": 2.6358100891113283,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.1375,
|
| 777 |
+
"grad_norm": 0.33104801177978516,
|
| 778 |
+
"learning_rate": 0.00011968958355901778,
|
| 779 |
+
"loss": 2.6341053009033204,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.13875,
|
| 784 |
+
"grad_norm": 0.3302455544471741,
|
| 785 |
+
"learning_rate": 0.00011960087207745023,
|
| 786 |
+
"loss": 2.659340667724609,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.14,
|
| 791 |
+
"grad_norm": 0.318013995885849,
|
| 792 |
+
"learning_rate": 0.00011951122374925103,
|
| 793 |
+
"loss": 2.6539737701416017,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.14125,
|
| 798 |
+
"grad_norm": 0.31688031554222107,
|
| 799 |
+
"learning_rate": 0.00011942064002871398,
|
| 800 |
+
"loss": 2.650745391845703,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.1425,
|
| 805 |
+
"grad_norm": 0.3218444883823395,
|
| 806 |
+
"learning_rate": 0.00011932912238530696,
|
| 807 |
+
"loss": 2.6293779373168946,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.14375,
|
| 812 |
+
"grad_norm": 0.31668025255203247,
|
| 813 |
+
"learning_rate": 0.0001192366723036482,
|
| 814 |
+
"loss": 2.652189254760742,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.145,
|
| 819 |
+
"grad_norm": 0.32894524931907654,
|
| 820 |
+
"learning_rate": 0.0001191432912834821,
|
| 821 |
+
"loss": 2.6034008026123048,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.14625,
|
| 826 |
+
"grad_norm": 0.326031357049942,
|
| 827 |
+
"learning_rate": 0.00011904898083965494,
|
| 828 |
+
"loss": 2.6356990814208983,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.1475,
|
| 833 |
+
"grad_norm": 0.3148091733455658,
|
| 834 |
+
"learning_rate": 0.00011895374250209033,
|
| 835 |
+
"loss": 2.6438148498535154,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.14875,
|
| 840 |
+
"grad_norm": 0.3154153823852539,
|
| 841 |
+
"learning_rate": 0.00011885757781576434,
|
| 842 |
+
"loss": 2.653242301940918,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.15,
|
| 847 |
+
"grad_norm": 0.31809449195861816,
|
| 848 |
+
"learning_rate": 0.00011876048834068046,
|
| 849 |
+
"loss": 2.6228126525878905,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.15125,
|
| 854 |
+
"grad_norm": 0.32725268602371216,
|
| 855 |
+
"learning_rate": 0.0001186624756518443,
|
| 856 |
+
"loss": 2.6216796875,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.1525,
|
| 861 |
+
"grad_norm": 0.32540032267570496,
|
| 862 |
+
"learning_rate": 0.00011856354133923805,
|
| 863 |
+
"loss": 2.67537841796875,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.15375,
|
| 868 |
+
"grad_norm": 0.3263508975505829,
|
| 869 |
+
"learning_rate": 0.00011846368700779467,
|
| 870 |
+
"loss": 2.6610176086425783,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.155,
|
| 875 |
+
"grad_norm": 0.3205776512622833,
|
| 876 |
+
"learning_rate": 0.00011836291427737183,
|
| 877 |
+
"loss": 2.6613521575927734,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.15625,
|
| 882 |
+
"grad_norm": 0.31028124690055847,
|
| 883 |
+
"learning_rate": 0.00011826122478272567,
|
| 884 |
+
"loss": 2.633769416809082,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.1575,
|
| 889 |
+
"grad_norm": 0.31673797965049744,
|
| 890 |
+
"learning_rate": 0.00011815862017348429,
|
| 891 |
+
"loss": 2.624924087524414,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.15875,
|
| 896 |
+
"grad_norm": 0.32373106479644775,
|
| 897 |
+
"learning_rate": 0.00011805510211412097,
|
| 898 |
+
"loss": 2.6462501525878905,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.16,
|
| 903 |
+
"grad_norm": 0.31725797057151794,
|
| 904 |
+
"learning_rate": 0.0001179506722839271,
|
| 905 |
+
"loss": 2.6365428924560548,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.16125,
|
| 910 |
+
"grad_norm": 0.3195420205593109,
|
| 911 |
+
"learning_rate": 0.00011784533237698511,
|
| 912 |
+
"loss": 2.6311481475830076,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.1625,
|
| 917 |
+
"grad_norm": 0.3341420888900757,
|
| 918 |
+
"learning_rate": 0.00011773908410214081,
|
| 919 |
+
"loss": 2.642291450500488,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.16375,
|
| 924 |
+
"grad_norm": 0.3230491876602173,
|
| 925 |
+
"learning_rate": 0.00011763192918297575,
|
| 926 |
+
"loss": 2.638113594055176,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.165,
|
| 931 |
+
"grad_norm": 0.3223067820072174,
|
| 932 |
+
"learning_rate": 0.0001175238693577793,
|
| 933 |
+
"loss": 2.6444271087646483,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.16625,
|
| 938 |
+
"grad_norm": 0.31934627890586853,
|
| 939 |
+
"learning_rate": 0.00011741490637952035,
|
| 940 |
+
"loss": 2.6657215118408204,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.1675,
|
| 945 |
+
"grad_norm": 0.3097170889377594,
|
| 946 |
+
"learning_rate": 0.00011730504201581893,
|
| 947 |
+
"loss": 2.645807647705078,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.16875,
|
| 952 |
+
"grad_norm": 0.32414084672927856,
|
| 953 |
+
"learning_rate": 0.00011719427804891757,
|
| 954 |
+
"loss": 2.641864776611328,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.17,
|
| 959 |
+
"grad_norm": 0.31383687257766724,
|
| 960 |
+
"learning_rate": 0.00011708261627565232,
|
| 961 |
+
"loss": 2.662236785888672,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.17125,
|
| 966 |
+
"grad_norm": 0.31501343846321106,
|
| 967 |
+
"learning_rate": 0.00011697005850742364,
|
| 968 |
+
"loss": 2.6557693481445312,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.1725,
|
| 973 |
+
"grad_norm": 0.31809887290000916,
|
| 974 |
+
"learning_rate": 0.00011685660657016701,
|
| 975 |
+
"loss": 2.6280593872070312,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.17375,
|
| 980 |
+
"grad_norm": 0.31885311007499695,
|
| 981 |
+
"learning_rate": 0.0001167422623043233,
|
| 982 |
+
"loss": 2.6564004898071287,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.175,
|
| 987 |
+
"grad_norm": 0.3105798065662384,
|
| 988 |
+
"learning_rate": 0.00011662702756480891,
|
| 989 |
+
"loss": 2.64355354309082,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.17625,
|
| 994 |
+
"grad_norm": 0.3361447751522064,
|
| 995 |
+
"learning_rate": 0.00011651090422098569,
|
| 996 |
+
"loss": 2.6594215393066407,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.1775,
|
| 1001 |
+
"grad_norm": 0.32253745198249817,
|
| 1002 |
+
"learning_rate": 0.00011639389415663065,
|
| 1003 |
+
"loss": 2.642239570617676,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.17875,
|
| 1008 |
+
"grad_norm": 0.32338932156562805,
|
| 1009 |
+
"learning_rate": 0.00011627599926990531,
|
| 1010 |
+
"loss": 2.6702959060668947,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.18,
|
| 1015 |
+
"grad_norm": 0.3116281032562256,
|
| 1016 |
+
"learning_rate": 0.00011615722147332501,
|
| 1017 |
+
"loss": 2.6370218276977537,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.18125,
|
| 1022 |
+
"grad_norm": 0.3282069265842438,
|
| 1023 |
+
"learning_rate": 0.00011603756269372781,
|
| 1024 |
+
"loss": 2.589012336730957,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.1825,
|
| 1029 |
+
"grad_norm": 0.32347872853279114,
|
| 1030 |
+
"learning_rate": 0.00011591702487224326,
|
| 1031 |
+
"loss": 2.638626480102539,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.18375,
|
| 1036 |
+
"grad_norm": 0.31963029503822327,
|
| 1037 |
+
"learning_rate": 0.0001157956099642609,
|
| 1038 |
+
"loss": 2.6150590896606447,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.185,
|
| 1043 |
+
"grad_norm": 0.31573331356048584,
|
| 1044 |
+
"learning_rate": 0.00011567331993939861,
|
| 1045 |
+
"loss": 2.6242300033569337,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.18625,
|
| 1050 |
+
"grad_norm": 0.318210631608963,
|
| 1051 |
+
"learning_rate": 0.00011555015678147051,
|
| 1052 |
+
"loss": 2.6236839294433594,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.1875,
|
| 1057 |
+
"grad_norm": 0.3299921751022339,
|
| 1058 |
+
"learning_rate": 0.0001154261224884549,
|
| 1059 |
+
"loss": 2.633551597595215,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.18875,
|
| 1064 |
+
"grad_norm": 0.32802239060401917,
|
| 1065 |
+
"learning_rate": 0.00011530121907246187,
|
| 1066 |
+
"loss": 2.650678253173828,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.19,
|
| 1071 |
+
"grad_norm": 0.3139156401157379,
|
| 1072 |
+
"learning_rate": 0.0001151754485597005,
|
| 1073 |
+
"loss": 2.6056631088256834,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.19125,
|
| 1078 |
+
"grad_norm": 0.320236474275589,
|
| 1079 |
+
"learning_rate": 0.00011504881299044619,
|
| 1080 |
+
"loss": 2.6355617523193358,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.1925,
|
| 1085 |
+
"grad_norm": 0.3379780054092407,
|
| 1086 |
+
"learning_rate": 0.00011492131441900742,
|
| 1087 |
+
"loss": 2.6405055999755858,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.19375,
|
| 1092 |
+
"grad_norm": 0.3395773470401764,
|
| 1093 |
+
"learning_rate": 0.00011479295491369245,
|
| 1094 |
+
"loss": 2.6217134475708006,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.195,
|
| 1099 |
+
"grad_norm": 0.33206456899642944,
|
| 1100 |
+
"learning_rate": 0.00011466373655677584,
|
| 1101 |
+
"loss": 2.6553268432617188,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.19625,
|
| 1106 |
+
"grad_norm": 0.3266463577747345,
|
| 1107 |
+
"learning_rate": 0.00011453366144446457,
|
| 1108 |
+
"loss": 2.615655517578125,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.1975,
|
| 1113 |
+
"grad_norm": 0.3166464567184448,
|
| 1114 |
+
"learning_rate": 0.0001144027316868641,
|
| 1115 |
+
"loss": 2.6240345001220704,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.19875,
|
| 1120 |
+
"grad_norm": 0.31986290216445923,
|
| 1121 |
+
"learning_rate": 0.00011427094940794416,
|
| 1122 |
+
"loss": 2.6230613708496096,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.2,
|
| 1127 |
+
"grad_norm": 0.3255802392959595,
|
| 1128 |
+
"learning_rate": 0.00011413831674550421,
|
| 1129 |
+
"loss": 2.6539276123046873,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.20125,
|
| 1134 |
+
"grad_norm": 0.3255312144756317,
|
| 1135 |
+
"learning_rate": 0.00011400483585113883,
|
| 1136 |
+
"loss": 2.6217121124267577,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.2025,
|
| 1141 |
+
"grad_norm": 0.3323643207550049,
|
| 1142 |
+
"learning_rate": 0.0001138705088902028,
|
| 1143 |
+
"loss": 2.652513885498047,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.20375,
|
| 1148 |
+
"grad_norm": 0.3227868974208832,
|
| 1149 |
+
"learning_rate": 0.00011373533804177592,
|
| 1150 |
+
"loss": 2.630014991760254,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.205,
|
| 1155 |
+
"grad_norm": 0.31701064109802246,
|
| 1156 |
+
"learning_rate": 0.00011359932549862779,
|
| 1157 |
+
"loss": 2.639967346191406,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.20625,
|
| 1162 |
+
"grad_norm": 0.3187071681022644,
|
| 1163 |
+
"learning_rate": 0.00011346247346718207,
|
| 1164 |
+
"loss": 2.6362884521484373,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.2075,
|
| 1169 |
+
"grad_norm": 0.31707099080085754,
|
| 1170 |
+
"learning_rate": 0.00011332478416748083,
|
| 1171 |
+
"loss": 2.649311065673828,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.20875,
|
| 1176 |
+
"grad_norm": 0.3297825753688812,
|
| 1177 |
+
"learning_rate": 0.00011318625983314848,
|
| 1178 |
+
"loss": 2.6421716690063475,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.21,
|
| 1183 |
+
"grad_norm": 0.3198815584182739,
|
| 1184 |
+
"learning_rate": 0.00011304690271135548,
|
| 1185 |
+
"loss": 2.633087730407715,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.21125,
|
| 1190 |
+
"grad_norm": 0.3226505219936371,
|
| 1191 |
+
"learning_rate": 0.00011290671506278205,
|
| 1192 |
+
"loss": 2.6442310333251955,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.2125,
|
| 1197 |
+
"grad_norm": 0.33370015025138855,
|
| 1198 |
+
"learning_rate": 0.00011276569916158123,
|
| 1199 |
+
"loss": 2.6304306030273437,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.21375,
|
| 1204 |
+
"grad_norm": 0.3307320773601532,
|
| 1205 |
+
"learning_rate": 0.0001126238572953423,
|
| 1206 |
+
"loss": 2.6353145599365235,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.215,
|
| 1211 |
+
"grad_norm": 0.31320619583129883,
|
| 1212 |
+
"learning_rate": 0.00011248119176505343,
|
| 1213 |
+
"loss": 2.6117172241210938,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.21625,
|
| 1218 |
+
"grad_norm": 0.3411354422569275,
|
| 1219 |
+
"learning_rate": 0.00011233770488506444,
|
| 1220 |
+
"loss": 2.6199378967285156,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.2175,
|
| 1225 |
+
"grad_norm": 0.3345658779144287,
|
| 1226 |
+
"learning_rate": 0.0001121933989830493,
|
| 1227 |
+
"loss": 2.617340850830078,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 0.21875,
|
| 1232 |
+
"grad_norm": 0.328173965215683,
|
| 1233 |
+
"learning_rate": 0.0001120482763999683,
|
| 1234 |
+
"loss": 2.646270751953125,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 0.22,
|
| 1239 |
+
"grad_norm": 0.31834596395492554,
|
| 1240 |
+
"learning_rate": 0.00011190233949003007,
|
| 1241 |
+
"loss": 2.6598697662353517,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.22125,
|
| 1246 |
+
"grad_norm": 0.32211023569107056,
|
| 1247 |
+
"learning_rate": 0.00011175559062065348,
|
| 1248 |
+
"loss": 2.617197036743164,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.2225,
|
| 1253 |
+
"grad_norm": 0.30770230293273926,
|
| 1254 |
+
"learning_rate": 0.00011160803217242911,
|
| 1255 |
+
"loss": 2.6376068115234377,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.22375,
|
| 1260 |
+
"grad_norm": 0.3243764042854309,
|
| 1261 |
+
"learning_rate": 0.00011145966653908078,
|
| 1262 |
+
"loss": 2.606427764892578,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.225,
|
| 1267 |
+
"grad_norm": 0.33548685908317566,
|
| 1268 |
+
"learning_rate": 0.00011131049612742655,
|
| 1269 |
+
"loss": 2.6384208679199217,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.22625,
|
| 1274 |
+
"grad_norm": 0.3262486159801483,
|
| 1275 |
+
"learning_rate": 0.00011116052335733979,
|
| 1276 |
+
"loss": 2.658290672302246,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 0.2275,
|
| 1281 |
+
"grad_norm": 0.31495559215545654,
|
| 1282 |
+
"learning_rate": 0.00011100975066170992,
|
| 1283 |
+
"loss": 2.662753105163574,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 0.22875,
|
| 1288 |
+
"grad_norm": 0.3250574469566345,
|
| 1289 |
+
"learning_rate": 0.00011085818048640288,
|
| 1290 |
+
"loss": 2.6388259887695313,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.23,
|
| 1295 |
+
"grad_norm": 0.34293144941329956,
|
| 1296 |
+
"learning_rate": 0.00011070581529022152,
|
| 1297 |
+
"loss": 2.6388187408447266,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.23125,
|
| 1302 |
+
"grad_norm": 0.31609639525413513,
|
| 1303 |
+
"learning_rate": 0.00011055265754486565,
|
| 1304 |
+
"loss": 2.637576675415039,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.2325,
|
| 1309 |
+
"grad_norm": 0.3181133270263672,
|
| 1310 |
+
"learning_rate": 0.00011039870973489204,
|
| 1311 |
+
"loss": 2.634903907775879,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 0.23375,
|
| 1316 |
+
"grad_norm": 0.3416786193847656,
|
| 1317 |
+
"learning_rate": 0.00011024397435767398,
|
| 1318 |
+
"loss": 2.616485023498535,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.235,
|
| 1323 |
+
"grad_norm": 0.315266489982605,
|
| 1324 |
+
"learning_rate": 0.00011008845392336087,
|
| 1325 |
+
"loss": 2.6373340606689455,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 0.23625,
|
| 1330 |
+
"grad_norm": 0.3316870927810669,
|
| 1331 |
+
"learning_rate": 0.0001099321509548375,
|
| 1332 |
+
"loss": 2.6363605499267577,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 0.2375,
|
| 1337 |
+
"grad_norm": 0.3230259418487549,
|
| 1338 |
+
"learning_rate": 0.00010977506798768303,
|
| 1339 |
+
"loss": 2.5958734512329102,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.23875,
|
| 1344 |
+
"grad_norm": 0.32272425293922424,
|
| 1345 |
+
"learning_rate": 0.00010961720757012995,
|
| 1346 |
+
"loss": 2.608958435058594,
|
| 1347 |
+
"step": 1910
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.24,
|
| 1351 |
+
"grad_norm": 0.3131502568721771,
|
| 1352 |
+
"learning_rate": 0.00010945857226302276,
|
| 1353 |
+
"loss": 2.6321544647216797,
|
| 1354 |
+
"step": 1920
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.24125,
|
| 1358 |
+
"grad_norm": 0.32143065333366394,
|
| 1359 |
+
"learning_rate": 0.00010929916463977628,
|
| 1360 |
+
"loss": 2.613364410400391,
|
| 1361 |
+
"step": 1930
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.2425,
|
| 1365 |
+
"grad_norm": 0.3152971565723419,
|
| 1366 |
+
"learning_rate": 0.00010913898728633408,
|
| 1367 |
+
"loss": 2.613265800476074,
|
| 1368 |
+
"step": 1940
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 0.24375,
|
| 1372 |
+
"grad_norm": 0.32848265767097473,
|
| 1373 |
+
"learning_rate": 0.00010897804280112643,
|
| 1374 |
+
"loss": 2.6013004302978517,
|
| 1375 |
+
"step": 1950
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 1.000875,
|
| 1379 |
+
"grad_norm": 0.3237718939781189,
|
| 1380 |
+
"learning_rate": 0.00010881633379502814,
|
| 1381 |
+
"loss": 2.8611122131347657,
|
| 1382 |
+
"step": 1960
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 1.002125,
|
| 1386 |
+
"grad_norm": 0.3280915915966034,
|
| 1387 |
+
"learning_rate": 0.00010865386289131632,
|
| 1388 |
+
"loss": 2.5412445068359375,
|
| 1389 |
+
"step": 1970
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 1.003375,
|
| 1393 |
+
"grad_norm": 0.33189550042152405,
|
| 1394 |
+
"learning_rate": 0.00010849063272562764,
|
| 1395 |
+
"loss": 2.559256362915039,
|
| 1396 |
+
"step": 1980
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 1.004625,
|
| 1400 |
+
"grad_norm": 0.3265272378921509,
|
| 1401 |
+
"learning_rate": 0.00010832664594591574,
|
| 1402 |
+
"loss": 2.5583423614501952,
|
| 1403 |
+
"step": 1990
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 1.005875,
|
| 1407 |
+
"grad_norm": 0.3453090488910675,
|
| 1408 |
+
"learning_rate": 0.00010816190521240819,
|
| 1409 |
+
"loss": 2.5712684631347655,
|
| 1410 |
+
"step": 2000
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 1.007125,
|
| 1414 |
+
"grad_norm": 0.3423366844654083,
|
| 1415 |
+
"learning_rate": 0.00010799641319756335,
|
| 1416 |
+
"loss": 2.5412336349487306,
|
| 1417 |
+
"step": 2010
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 1.008375,
|
| 1421 |
+
"grad_norm": 0.32097378373146057,
|
| 1422 |
+
"learning_rate": 0.00010783017258602704,
|
| 1423 |
+
"loss": 2.5253084182739256,
|
| 1424 |
+
"step": 2020
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 1.009625,
|
| 1428 |
+
"grad_norm": 0.3252958357334137,
|
| 1429 |
+
"learning_rate": 0.00010766318607458898,
|
| 1430 |
+
"loss": 2.5738031387329103,
|
| 1431 |
+
"step": 2030
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 1.010875,
|
| 1435 |
+
"grad_norm": 0.3372173011302948,
|
| 1436 |
+
"learning_rate": 0.00010749545637213897,
|
| 1437 |
+
"loss": 2.54388370513916,
|
| 1438 |
+
"step": 2040
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 1.012125,
|
| 1442 |
+
"grad_norm": 0.33359599113464355,
|
| 1443 |
+
"learning_rate": 0.00010732698619962306,
|
| 1444 |
+
"loss": 2.55248908996582,
|
| 1445 |
+
"step": 2050
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 1.013375,
|
| 1449 |
+
"grad_norm": 0.34591928124427795,
|
| 1450 |
+
"learning_rate": 0.00010715777828999937,
|
| 1451 |
+
"loss": 2.5376352310180663,
|
| 1452 |
+
"step": 2060
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 1.014625,
|
| 1456 |
+
"grad_norm": 0.35322073101997375,
|
| 1457 |
+
"learning_rate": 0.00010698783538819372,
|
| 1458 |
+
"loss": 2.534122085571289,
|
| 1459 |
+
"step": 2070
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 1.015875,
|
| 1463 |
+
"grad_norm": 0.3539016544818878,
|
| 1464 |
+
"learning_rate": 0.00010681716025105512,
|
| 1465 |
+
"loss": 2.492664337158203,
|
| 1466 |
+
"step": 2080
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 1.017125,
|
| 1470 |
+
"grad_norm": 0.3429170548915863,
|
| 1471 |
+
"learning_rate": 0.00010664575564731107,
|
| 1472 |
+
"loss": 2.5008804321289064,
|
| 1473 |
+
"step": 2090
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 1.018375,
|
| 1477 |
+
"grad_norm": 0.3576091229915619,
|
| 1478 |
+
"learning_rate": 0.00010647362435752263,
|
| 1479 |
+
"loss": 2.5176633834838866,
|
| 1480 |
+
"step": 2100
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 1.019625,
|
| 1484 |
+
"grad_norm": 0.3297135829925537,
|
| 1485 |
+
"learning_rate": 0.00010630076917403929,
|
| 1486 |
+
"loss": 2.500911331176758,
|
| 1487 |
+
"step": 2110
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 1.020875,
|
| 1491 |
+
"grad_norm": 0.33164292573928833,
|
| 1492 |
+
"learning_rate": 0.00010612719290095374,
|
| 1493 |
+
"loss": 2.513214111328125,
|
| 1494 |
+
"step": 2120
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 1.022125,
|
| 1498 |
+
"grad_norm": 0.32680895924568176,
|
| 1499 |
+
"learning_rate": 0.00010595289835405624,
|
| 1500 |
+
"loss": 2.501193809509277,
|
| 1501 |
+
"step": 2130
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 1.023375,
|
| 1505 |
+
"grad_norm": 0.33465543389320374,
|
| 1506 |
+
"learning_rate": 0.00010577788836078916,
|
| 1507 |
+
"loss": 2.4999351501464844,
|
| 1508 |
+
"step": 2140
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 1.024625,
|
| 1512 |
+
"grad_norm": 0.3344171941280365,
|
| 1513 |
+
"learning_rate": 0.00010560216576020092,
|
| 1514 |
+
"loss": 2.4867813110351564,
|
| 1515 |
+
"step": 2150
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 1.025875,
|
| 1519 |
+
"grad_norm": 0.34607550501823425,
|
| 1520 |
+
"learning_rate": 0.00010542573340289998,
|
| 1521 |
+
"loss": 2.503824234008789,
|
| 1522 |
+
"step": 2160
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 1.027125,
|
| 1526 |
+
"grad_norm": 0.33892592787742615,
|
| 1527 |
+
"learning_rate": 0.00010524859415100871,
|
| 1528 |
+
"loss": 2.4990135192871095,
|
| 1529 |
+
"step": 2170
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 1.028375,
|
| 1533 |
+
"grad_norm": 0.3448082208633423,
|
| 1534 |
+
"learning_rate": 0.00010507075087811677,
|
| 1535 |
+
"loss": 2.4324840545654296,
|
| 1536 |
+
"step": 2180
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 1.029625,
|
| 1540 |
+
"grad_norm": 0.3321894407272339,
|
| 1541 |
+
"learning_rate": 0.00010489220646923464,
|
| 1542 |
+
"loss": 2.4842708587646483,
|
| 1543 |
+
"step": 2190
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 1.030875,
|
| 1547 |
+
"grad_norm": 0.3443576395511627,
|
| 1548 |
+
"learning_rate": 0.0001047129638207468,
|
| 1549 |
+
"loss": 2.485816764831543,
|
| 1550 |
+
"step": 2200
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 1.032125,
|
| 1554 |
+
"grad_norm": 0.3381134271621704,
|
| 1555 |
+
"learning_rate": 0.00010453302584036468,
|
| 1556 |
+
"loss": 2.4841537475585938,
|
| 1557 |
+
"step": 2210
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 1.033375,
|
| 1561 |
+
"grad_norm": 0.3401469588279724,
|
| 1562 |
+
"learning_rate": 0.00010435239544707952,
|
| 1563 |
+
"loss": 2.48382453918457,
|
| 1564 |
+
"step": 2220
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 1.034625,
|
| 1568 |
+
"grad_norm": 0.35364025831222534,
|
| 1569 |
+
"learning_rate": 0.00010417107557111507,
|
| 1570 |
+
"loss": 2.4872058868408202,
|
| 1571 |
+
"step": 2230
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 1.035875,
|
| 1575 |
+
"grad_norm": 0.3584776818752289,
|
| 1576 |
+
"learning_rate": 0.00010398906915388,
|
| 1577 |
+
"loss": 2.455089569091797,
|
| 1578 |
+
"step": 2240
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 1.037125,
|
| 1582 |
+
"grad_norm": 0.3385666608810425,
|
| 1583 |
+
"learning_rate": 0.00010380637914792015,
|
| 1584 |
+
"loss": 2.4457998275756836,
|
| 1585 |
+
"step": 2250
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 1.038375,
|
| 1589 |
+
"grad_norm": 0.3520835340023041,
|
| 1590 |
+
"learning_rate": 0.00010362300851687071,
|
| 1591 |
+
"loss": 2.479095458984375,
|
| 1592 |
+
"step": 2260
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 1.039625,
|
| 1596 |
+
"grad_norm": 0.34799060225486755,
|
| 1597 |
+
"learning_rate": 0.00010343896023540814,
|
| 1598 |
+
"loss": 2.4659198760986327,
|
| 1599 |
+
"step": 2270
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 1.040875,
|
| 1603 |
+
"grad_norm": 0.35186630487442017,
|
| 1604 |
+
"learning_rate": 0.00010325423728920182,
|
| 1605 |
+
"loss": 2.4467798233032227,
|
| 1606 |
+
"step": 2280
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 1.042125,
|
| 1610 |
+
"grad_norm": 0.3423445224761963,
|
| 1611 |
+
"learning_rate": 0.00010306884267486574,
|
| 1612 |
+
"loss": 2.4702438354492187,
|
| 1613 |
+
"step": 2290
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 1.043375,
|
| 1617 |
+
"grad_norm": 0.3398495018482208,
|
| 1618 |
+
"learning_rate": 0.00010288277939990981,
|
| 1619 |
+
"loss": 2.471152496337891,
|
| 1620 |
+
"step": 2300
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 1.044625,
|
| 1624 |
+
"grad_norm": 0.34717217087745667,
|
| 1625 |
+
"learning_rate": 0.00010269605048269109,
|
| 1626 |
+
"loss": 2.4720317840576174,
|
| 1627 |
+
"step": 2310
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 1.045875,
|
| 1631 |
+
"grad_norm": 0.34331125020980835,
|
| 1632 |
+
"learning_rate": 0.00010250865895236482,
|
| 1633 |
+
"loss": 2.4562469482421876,
|
| 1634 |
+
"step": 2320
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 1.047125,
|
| 1638 |
+
"grad_norm": 0.35022589564323425,
|
| 1639 |
+
"learning_rate": 0.00010232060784883528,
|
| 1640 |
+
"loss": 2.461803436279297,
|
| 1641 |
+
"step": 2330
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 1.048375,
|
| 1645 |
+
"grad_norm": 0.3725920617580414,
|
| 1646 |
+
"learning_rate": 0.00010213190022270653,
|
| 1647 |
+
"loss": 2.4350805282592773,
|
| 1648 |
+
"step": 2340
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 1.049625,
|
| 1652 |
+
"grad_norm": 0.3634240925312042,
|
| 1653 |
+
"learning_rate": 0.00010194253913523282,
|
| 1654 |
+
"loss": 2.454206848144531,
|
| 1655 |
+
"step": 2350
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 1.050875,
|
| 1659 |
+
"grad_norm": 0.35172227025032043,
|
| 1660 |
+
"learning_rate": 0.000101752527658269,
|
| 1661 |
+
"loss": 2.4318115234375,
|
| 1662 |
+
"step": 2360
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 1.052125,
|
| 1666 |
+
"grad_norm": 0.35827362537384033,
|
| 1667 |
+
"learning_rate": 0.00010156186887422071,
|
| 1668 |
+
"loss": 2.4692001342773438,
|
| 1669 |
+
"step": 2370
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 1.053375,
|
| 1673 |
+
"grad_norm": 0.36834755539894104,
|
| 1674 |
+
"learning_rate": 0.00010137056587599428,
|
| 1675 |
+
"loss": 2.4683910369873048,
|
| 1676 |
+
"step": 2380
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 1.054625,
|
| 1680 |
+
"grad_norm": 0.3573245108127594,
|
| 1681 |
+
"learning_rate": 0.00010117862176694666,
|
| 1682 |
+
"loss": 2.4428688049316407,
|
| 1683 |
+
"step": 2390
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 1.055875,
|
| 1687 |
+
"grad_norm": 0.33202221989631653,
|
| 1688 |
+
"learning_rate": 0.00010098603966083503,
|
| 1689 |
+
"loss": 2.4585454940795897,
|
| 1690 |
+
"step": 2400
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 1.057125,
|
| 1694 |
+
"grad_norm": 0.35598650574684143,
|
| 1695 |
+
"learning_rate": 0.00010079282268176628,
|
| 1696 |
+
"loss": 2.4740036010742186,
|
| 1697 |
+
"step": 2410
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 1.058375,
|
| 1701 |
+
"grad_norm": 0.36041730642318726,
|
| 1702 |
+
"learning_rate": 0.00010059897396414633,
|
| 1703 |
+
"loss": 2.4598981857299806,
|
| 1704 |
+
"step": 2420
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 1.059625,
|
| 1708 |
+
"grad_norm": 0.3481718599796295,
|
| 1709 |
+
"learning_rate": 0.00010040449665262931,
|
| 1710 |
+
"loss": 2.4539608001708983,
|
| 1711 |
+
"step": 2430
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 1.060875,
|
| 1715 |
+
"grad_norm": 0.3672044277191162,
|
| 1716 |
+
"learning_rate": 0.00010020939390206654,
|
| 1717 |
+
"loss": 2.433728790283203,
|
| 1718 |
+
"step": 2440
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 1.062125,
|
| 1722 |
+
"grad_norm": 0.35078802704811096,
|
| 1723 |
+
"learning_rate": 0.00010001366887745531,
|
| 1724 |
+
"loss": 2.454706573486328,
|
| 1725 |
+
"step": 2450
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 1.063375,
|
| 1729 |
+
"grad_norm": 0.36530086398124695,
|
| 1730 |
+
"learning_rate": 9.981732475388758e-05,
|
| 1731 |
+
"loss": 2.4748252868652343,
|
| 1732 |
+
"step": 2460
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 1.064625,
|
| 1736 |
+
"grad_norm": 0.3578907251358032,
|
| 1737 |
+
"learning_rate": 9.962036471649851e-05,
|
| 1738 |
+
"loss": 2.480423355102539,
|
| 1739 |
+
"step": 2470
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 1.065875,
|
| 1743 |
+
"grad_norm": 0.370403528213501,
|
| 1744 |
+
"learning_rate": 9.942279196041466e-05,
|
| 1745 |
+
"loss": 2.480521011352539,
|
| 1746 |
+
"step": 2480
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 1.067125,
|
| 1750 |
+
"grad_norm": 0.36263224482536316,
|
| 1751 |
+
"learning_rate": 9.922460969070231e-05,
|
| 1752 |
+
"loss": 2.4786655426025392,
|
| 1753 |
+
"step": 2490
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 1.068375,
|
| 1757 |
+
"grad_norm": 0.35858920216560364,
|
| 1758 |
+
"learning_rate": 9.902582112231533e-05,
|
| 1759 |
+
"loss": 2.461780548095703,
|
| 1760 |
+
"step": 2500
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 1.069625,
|
| 1764 |
+
"grad_norm": 0.35088691115379333,
|
| 1765 |
+
"learning_rate": 9.882642948004314e-05,
|
| 1766 |
+
"loss": 2.4797664642333985,
|
| 1767 |
+
"step": 2510
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 1.070875,
|
| 1771 |
+
"grad_norm": 0.36078205704689026,
|
| 1772 |
+
"learning_rate": 9.862643799845839e-05,
|
| 1773 |
+
"loss": 2.4529985427856444,
|
| 1774 |
+
"step": 2520
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 1.072125,
|
| 1778 |
+
"grad_norm": 0.35207876563072205,
|
| 1779 |
+
"learning_rate": 9.842584992186434e-05,
|
| 1780 |
+
"loss": 2.4753444671630858,
|
| 1781 |
+
"step": 2530
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 1.073375,
|
| 1785 |
+
"grad_norm": 0.36091098189353943,
|
| 1786 |
+
"learning_rate": 9.822466850424243e-05,
|
| 1787 |
+
"loss": 2.4327056884765623,
|
| 1788 |
+
"step": 2540
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 1.074625,
|
| 1792 |
+
"grad_norm": 0.3577967584133148,
|
| 1793 |
+
"learning_rate": 9.802289700919933e-05,
|
| 1794 |
+
"loss": 2.461964416503906,
|
| 1795 |
+
"step": 2550
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 1.075875,
|
| 1799 |
+
"grad_norm": 0.36178645491600037,
|
| 1800 |
+
"learning_rate": 9.782053870991414e-05,
|
| 1801 |
+
"loss": 2.4669708251953124,
|
| 1802 |
+
"step": 2560
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 1.077125,
|
| 1806 |
+
"grad_norm": 0.3427974581718445,
|
| 1807 |
+
"learning_rate": 9.761759688908519e-05,
|
| 1808 |
+
"loss": 2.4416053771972654,
|
| 1809 |
+
"step": 2570
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 1.078375,
|
| 1813 |
+
"grad_norm": 0.3656075894832611,
|
| 1814 |
+
"learning_rate": 9.741407483887678e-05,
|
| 1815 |
+
"loss": 2.4402462005615235,
|
| 1816 |
+
"step": 2580
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 1.079625,
|
| 1820 |
+
"grad_norm": 0.37002095580101013,
|
| 1821 |
+
"learning_rate": 9.720997586086587e-05,
|
| 1822 |
+
"loss": 2.451791191101074,
|
| 1823 |
+
"step": 2590
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 1.080875,
|
| 1827 |
+
"grad_norm": 0.3515098989009857,
|
| 1828 |
+
"learning_rate": 9.700530326598842e-05,
|
| 1829 |
+
"loss": 2.459187889099121,
|
| 1830 |
+
"step": 2600
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 1.082125,
|
| 1834 |
+
"grad_norm": 0.3765217959880829,
|
| 1835 |
+
"learning_rate": 9.680006037448575e-05,
|
| 1836 |
+
"loss": 2.4384769439697265,
|
| 1837 |
+
"step": 2610
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 1.083375,
|
| 1841 |
+
"grad_norm": 0.5566070675849915,
|
| 1842 |
+
"learning_rate": 9.659425051585065e-05,
|
| 1843 |
+
"loss": 2.4481531143188477,
|
| 1844 |
+
"step": 2620
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 1.084625,
|
| 1848 |
+
"grad_norm": 0.3657555878162384,
|
| 1849 |
+
"learning_rate": 9.638787702877333e-05,
|
| 1850 |
+
"loss": 2.470143508911133,
|
| 1851 |
+
"step": 2630
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 1.085875,
|
| 1855 |
+
"grad_norm": 0.35306495428085327,
|
| 1856 |
+
"learning_rate": 9.618094326108734e-05,
|
| 1857 |
+
"loss": 2.4623140335083007,
|
| 1858 |
+
"step": 2640
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 1.087125,
|
| 1862 |
+
"grad_norm": 0.35708507895469666,
|
| 1863 |
+
"learning_rate": 9.597345256971521e-05,
|
| 1864 |
+
"loss": 2.4393037796020507,
|
| 1865 |
+
"step": 2650
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 1.088375,
|
| 1869 |
+
"grad_norm": 0.36429449915885925,
|
| 1870 |
+
"learning_rate": 9.576540832061398e-05,
|
| 1871 |
+
"loss": 2.4460866928100584,
|
| 1872 |
+
"step": 2660
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 1.089625,
|
| 1876 |
+
"grad_norm": 0.3617342710494995,
|
| 1877 |
+
"learning_rate": 9.555681388872065e-05,
|
| 1878 |
+
"loss": 2.476423454284668,
|
| 1879 |
+
"step": 2670
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 1.090875,
|
| 1883 |
+
"grad_norm": 0.3526591360569,
|
| 1884 |
+
"learning_rate": 9.534767265789737e-05,
|
| 1885 |
+
"loss": 2.460892105102539,
|
| 1886 |
+
"step": 2680
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 1.092125,
|
| 1890 |
+
"grad_norm": 0.3697713613510132,
|
| 1891 |
+
"learning_rate": 9.51379880208766e-05,
|
| 1892 |
+
"loss": 2.46860294342041,
|
| 1893 |
+
"step": 2690
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 1.093375,
|
| 1897 |
+
"grad_norm": 0.37604451179504395,
|
| 1898 |
+
"learning_rate": 9.492776337920603e-05,
|
| 1899 |
+
"loss": 2.465809631347656,
|
| 1900 |
+
"step": 2700
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 1.094625,
|
| 1904 |
+
"grad_norm": 0.37269482016563416,
|
| 1905 |
+
"learning_rate": 9.471700214319343e-05,
|
| 1906 |
+
"loss": 2.4291683197021485,
|
| 1907 |
+
"step": 2710
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 1.095875,
|
| 1911 |
+
"grad_norm": 0.37273484468460083,
|
| 1912 |
+
"learning_rate": 9.45057077318513e-05,
|
| 1913 |
+
"loss": 2.447264862060547,
|
| 1914 |
+
"step": 2720
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 1.097125,
|
| 1918 |
+
"grad_norm": 0.3633696734905243,
|
| 1919 |
+
"learning_rate": 9.429388357284143e-05,
|
| 1920 |
+
"loss": 2.471749114990234,
|
| 1921 |
+
"step": 2730
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 1.098375,
|
| 1925 |
+
"grad_norm": 0.35682767629623413,
|
| 1926 |
+
"learning_rate": 9.40815331024193e-05,
|
| 1927 |
+
"loss": 2.42556209564209,
|
| 1928 |
+
"step": 2740
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 1.099625,
|
| 1932 |
+
"grad_norm": 0.3471936285495758,
|
| 1933 |
+
"learning_rate": 9.386865976537827e-05,
|
| 1934 |
+
"loss": 2.446389007568359,
|
| 1935 |
+
"step": 2750
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 1.100875,
|
| 1939 |
+
"grad_norm": 0.38089418411254883,
|
| 1940 |
+
"learning_rate": 9.365526701499384e-05,
|
| 1941 |
+
"loss": 2.4501571655273438,
|
| 1942 |
+
"step": 2760
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 1.102125,
|
| 1946 |
+
"grad_norm": 0.3654205799102783,
|
| 1947 |
+
"learning_rate": 9.344135831296749e-05,
|
| 1948 |
+
"loss": 2.439041519165039,
|
| 1949 |
+
"step": 2770
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 1.103375,
|
| 1953 |
+
"grad_norm": 0.3512708842754364,
|
| 1954 |
+
"learning_rate": 9.322693712937054e-05,
|
| 1955 |
+
"loss": 2.4336933135986327,
|
| 1956 |
+
"step": 2780
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 1.104625,
|
| 1960 |
+
"grad_norm": 0.36569294333457947,
|
| 1961 |
+
"learning_rate": 9.301200694258795e-05,
|
| 1962 |
+
"loss": 2.444048309326172,
|
| 1963 |
+
"step": 2790
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 1.105875,
|
| 1967 |
+
"grad_norm": 0.36901962757110596,
|
| 1968 |
+
"learning_rate": 9.279657123926178e-05,
|
| 1969 |
+
"loss": 2.4316547393798826,
|
| 1970 |
+
"step": 2800
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 1.107125,
|
| 1974 |
+
"grad_norm": 0.36593225598335266,
|
| 1975 |
+
"learning_rate": 9.25806335142348e-05,
|
| 1976 |
+
"loss": 2.4314062118530275,
|
| 1977 |
+
"step": 2810
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 1.108375,
|
| 1981 |
+
"grad_norm": 0.371039479970932,
|
| 1982 |
+
"learning_rate": 9.236419727049352e-05,
|
| 1983 |
+
"loss": 2.4478275299072267,
|
| 1984 |
+
"step": 2820
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 1.109625,
|
| 1988 |
+
"grad_norm": 0.3607841730117798,
|
| 1989 |
+
"learning_rate": 9.214726601911162e-05,
|
| 1990 |
+
"loss": 2.471347999572754,
|
| 1991 |
+
"step": 2830
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 1.110875,
|
| 1995 |
+
"grad_norm": 0.35733747482299805,
|
| 1996 |
+
"learning_rate": 9.192984327919289e-05,
|
| 1997 |
+
"loss": 2.4454570770263673,
|
| 1998 |
+
"step": 2840
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 1.112125,
|
| 2002 |
+
"grad_norm": 0.3512793183326721,
|
| 2003 |
+
"learning_rate": 9.171193257781413e-05,
|
| 2004 |
+
"loss": 2.4474578857421876,
|
| 2005 |
+
"step": 2850
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 1.113375,
|
| 2009 |
+
"grad_norm": 0.3591439127922058,
|
| 2010 |
+
"learning_rate": 9.149353744996798e-05,
|
| 2011 |
+
"loss": 2.3968666076660154,
|
| 2012 |
+
"step": 2860
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 1.114625,
|
| 2016 |
+
"grad_norm": 0.37512722611427307,
|
| 2017 |
+
"learning_rate": 9.127466143850551e-05,
|
| 2018 |
+
"loss": 2.4625476837158202,
|
| 2019 |
+
"step": 2870
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 1.115875,
|
| 2023 |
+
"grad_norm": 0.3683817982673645,
|
| 2024 |
+
"learning_rate": 9.105530809407877e-05,
|
| 2025 |
+
"loss": 2.4239782333374023,
|
| 2026 |
+
"step": 2880
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 1.117125,
|
| 2030 |
+
"grad_norm": 0.350392609834671,
|
| 2031 |
+
"learning_rate": 9.08354809750833e-05,
|
| 2032 |
+
"loss": 2.4604770660400392,
|
| 2033 |
+
"step": 2890
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 1.118375,
|
| 2037 |
+
"grad_norm": 0.3627133071422577,
|
| 2038 |
+
"learning_rate": 9.061518364760018e-05,
|
| 2039 |
+
"loss": 2.4404422760009767,
|
| 2040 |
+
"step": 2900
|
| 2041 |
+
}
|
| 2042 |
+
],
|
| 2043 |
+
"logging_steps": 10,
|
| 2044 |
+
"max_steps": 8000,
|
| 2045 |
+
"num_input_tokens_seen": 0,
|
| 2046 |
+
"num_train_epochs": 9223372036854775807,
|
| 2047 |
+
"save_steps": 100,
|
| 2048 |
+
"stateful_callbacks": {
|
| 2049 |
+
"TrainerControl": {
|
| 2050 |
+
"args": {
|
| 2051 |
+
"should_epoch_stop": false,
|
| 2052 |
+
"should_evaluate": false,
|
| 2053 |
+
"should_log": false,
|
| 2054 |
+
"should_save": true,
|
| 2055 |
+
"should_training_stop": false
|
| 2056 |
+
},
|
| 2057 |
+
"attributes": {}
|
| 2058 |
+
}
|
| 2059 |
+
},
|
| 2060 |
+
"total_flos": 3.8847708770441626e+17,
|
| 2061 |
+
"train_batch_size": 4,
|
| 2062 |
+
"trial_name": null,
|
| 2063 |
+
"trial_params": null
|
| 2064 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b704d3db9a933825d1f626a6c46e24e399262779ae9a310bc6f83e9af2d52373
|
| 3 |
+
size 5265
|