Upload folder using huggingface_hub
Browse files- config.json +43 -0
- mm_projector.bin +3 -0
- trainer_state.json +1421 -0
config.json
ADDED
|
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "/model_zoo/SeaLLM-7B-v2",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"MistralForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"bert_type": "qformer_layer:12",
|
| 8 |
+
"bos_token_id": 1,
|
| 9 |
+
"compress_type": "mean",
|
| 10 |
+
"eos_token_id": 2,
|
| 11 |
+
"freeze_mm_mlp_adapter": false,
|
| 12 |
+
"hidden_act": "silu",
|
| 13 |
+
"hidden_size": 4096,
|
| 14 |
+
"image_aspect_ratio": "square",
|
| 15 |
+
"image_grid_pinpoints": null,
|
| 16 |
+
"image_processor": "./llamavid/processor/intern-vit",
|
| 17 |
+
"initializer_range": 0.02,
|
| 18 |
+
"intermediate_size": 14336,
|
| 19 |
+
"max_position_embeddings": 32768,
|
| 20 |
+
"max_token": 2048,
|
| 21 |
+
"mm_hidden_size": 1024,
|
| 22 |
+
"mm_projector_type": "mlp2x_gelu",
|
| 23 |
+
"mm_use_im_patch_token": false,
|
| 24 |
+
"mm_use_im_start_end": false,
|
| 25 |
+
"mm_vision_select_feature": "patch",
|
| 26 |
+
"mm_vision_select_layer": -1,
|
| 27 |
+
"mm_vision_tower": "/model_zoo/intern-vit",
|
| 28 |
+
"model_type": "mistral",
|
| 29 |
+
"num_attention_heads": 32,
|
| 30 |
+
"num_hidden_layers": 32,
|
| 31 |
+
"num_key_value_heads": 8,
|
| 32 |
+
"num_query": 32,
|
| 33 |
+
"rms_norm_eps": 1e-05,
|
| 34 |
+
"rope_theta": 10000.0,
|
| 35 |
+
"sliding_window": 4096,
|
| 36 |
+
"tie_word_embeddings": false,
|
| 37 |
+
"torch_dtype": "bfloat16",
|
| 38 |
+
"transformers_version": "4.38.2",
|
| 39 |
+
"tune_mm_mlp_adapter": true,
|
| 40 |
+
"use_cache": false,
|
| 41 |
+
"use_mm_proj": true,
|
| 42 |
+
"vocab_size": 48384
|
| 43 |
+
}
|
mm_projector.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b24e5cb2390683fb253779422dd9b68a85e334eda3b72c209638f0880fd5d26a
|
| 3 |
+
size 469474367
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1421 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.6099885627144491,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 200,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": false,
|
| 9 |
+
"is_world_process_zero": false,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"grad_norm": 26.258708272339597,
|
| 14 |
+
"learning_rate": 0.0001,
|
| 15 |
+
"loss": 6.0469,
|
| 16 |
+
"step": 1
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.01,
|
| 20 |
+
"grad_norm": 20.031528084144362,
|
| 21 |
+
"learning_rate": 0.0002,
|
| 22 |
+
"loss": 6.0524,
|
| 23 |
+
"step": 2
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.01,
|
| 27 |
+
"grad_norm": 9.323112879274246,
|
| 28 |
+
"learning_rate": 0.0003,
|
| 29 |
+
"loss": 6.0052,
|
| 30 |
+
"step": 3
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.01,
|
| 34 |
+
"grad_norm": 4.147754138762137,
|
| 35 |
+
"learning_rate": 0.0004,
|
| 36 |
+
"loss": 5.7581,
|
| 37 |
+
"step": 4
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.02,
|
| 41 |
+
"grad_norm": 1.4117904450662409,
|
| 42 |
+
"learning_rate": 0.0005,
|
| 43 |
+
"loss": 5.5959,
|
| 44 |
+
"step": 5
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.02,
|
| 48 |
+
"grad_norm": 1.450165632053372,
|
| 49 |
+
"learning_rate": 0.0006,
|
| 50 |
+
"loss": 5.3926,
|
| 51 |
+
"step": 6
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.02,
|
| 55 |
+
"grad_norm": 1.3379034522061206,
|
| 56 |
+
"learning_rate": 0.0007,
|
| 57 |
+
"loss": 5.2432,
|
| 58 |
+
"step": 7
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.02,
|
| 62 |
+
"grad_norm": 1.3149419664848485,
|
| 63 |
+
"learning_rate": 0.0008,
|
| 64 |
+
"loss": 4.9048,
|
| 65 |
+
"step": 8
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.03,
|
| 69 |
+
"grad_norm": 1.5265629850853388,
|
| 70 |
+
"learning_rate": 0.0009000000000000001,
|
| 71 |
+
"loss": 4.8456,
|
| 72 |
+
"step": 9
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.03,
|
| 76 |
+
"grad_norm": 1.2248009557702129,
|
| 77 |
+
"learning_rate": 0.001,
|
| 78 |
+
"loss": 4.4037,
|
| 79 |
+
"step": 10
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.03,
|
| 83 |
+
"grad_norm": 1.5753342093655753,
|
| 84 |
+
"learning_rate": 0.0009999754462587395,
|
| 85 |
+
"loss": 3.7762,
|
| 86 |
+
"step": 11
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.04,
|
| 90 |
+
"grad_norm": 1.0959671581261972,
|
| 91 |
+
"learning_rate": 0.0009999017874465026,
|
| 92 |
+
"loss": 3.5411,
|
| 93 |
+
"step": 12
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.04,
|
| 97 |
+
"grad_norm": 0.8383646382276924,
|
| 98 |
+
"learning_rate": 0.0009997790307976872,
|
| 99 |
+
"loss": 3.4374,
|
| 100 |
+
"step": 13
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.04,
|
| 104 |
+
"grad_norm": 0.6879126623626883,
|
| 105 |
+
"learning_rate": 0.0009996071883688333,
|
| 106 |
+
"loss": 3.3495,
|
| 107 |
+
"step": 14
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.05,
|
| 111 |
+
"grad_norm": 0.48057277440632934,
|
| 112 |
+
"learning_rate": 0.000999386277037439,
|
| 113 |
+
"loss": 3.3236,
|
| 114 |
+
"step": 15
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.05,
|
| 118 |
+
"grad_norm": 0.5193770648047042,
|
| 119 |
+
"learning_rate": 0.0009991163185003027,
|
| 120 |
+
"loss": 3.194,
|
| 121 |
+
"step": 16
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.05,
|
| 125 |
+
"grad_norm": 0.34046016814570407,
|
| 126 |
+
"learning_rate": 0.0009987973392713932,
|
| 127 |
+
"loss": 3.008,
|
| 128 |
+
"step": 17
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.05,
|
| 132 |
+
"grad_norm": 0.2969743503479429,
|
| 133 |
+
"learning_rate": 0.0009984293706792437,
|
| 134 |
+
"loss": 3.0682,
|
| 135 |
+
"step": 18
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.06,
|
| 139 |
+
"grad_norm": 0.22754131045377113,
|
| 140 |
+
"learning_rate": 0.0009980124488638772,
|
| 141 |
+
"loss": 2.92,
|
| 142 |
+
"step": 19
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.06,
|
| 146 |
+
"grad_norm": 0.15880784203685044,
|
| 147 |
+
"learning_rate": 0.000997546614773255,
|
| 148 |
+
"loss": 2.881,
|
| 149 |
+
"step": 20
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.06,
|
| 153 |
+
"grad_norm": 0.10804346892890891,
|
| 154 |
+
"learning_rate": 0.0009970319141592559,
|
| 155 |
+
"loss": 2.8386,
|
| 156 |
+
"step": 21
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.07,
|
| 160 |
+
"grad_norm": 0.0789962600177391,
|
| 161 |
+
"learning_rate": 0.0009964683975731828,
|
| 162 |
+
"loss": 2.8704,
|
| 163 |
+
"step": 22
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.07,
|
| 167 |
+
"grad_norm": 0.0678766718481922,
|
| 168 |
+
"learning_rate": 0.0009958561203607973,
|
| 169 |
+
"loss": 2.8326,
|
| 170 |
+
"step": 23
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.07,
|
| 174 |
+
"grad_norm": 0.15129601777209895,
|
| 175 |
+
"learning_rate": 0.000995195142656885,
|
| 176 |
+
"loss": 2.8444,
|
| 177 |
+
"step": 24
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.08,
|
| 181 |
+
"grad_norm": 0.23876681297628152,
|
| 182 |
+
"learning_rate": 0.0009944855293793475,
|
| 183 |
+
"loss": 2.8809,
|
| 184 |
+
"step": 25
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.08,
|
| 188 |
+
"grad_norm": 0.2458680275413572,
|
| 189 |
+
"learning_rate": 0.0009937273502228282,
|
| 190 |
+
"loss": 2.8308,
|
| 191 |
+
"step": 26
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.08,
|
| 195 |
+
"grad_norm": 0.2646415320413548,
|
| 196 |
+
"learning_rate": 0.0009929206796518663,
|
| 197 |
+
"loss": 2.8296,
|
| 198 |
+
"step": 27
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.09,
|
| 202 |
+
"grad_norm": 0.1855177204894958,
|
| 203 |
+
"learning_rate": 0.0009920655968935837,
|
| 204 |
+
"loss": 2.8302,
|
| 205 |
+
"step": 28
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.09,
|
| 209 |
+
"grad_norm": 0.1692458314027983,
|
| 210 |
+
"learning_rate": 0.0009911621859299041,
|
| 211 |
+
"loss": 2.7798,
|
| 212 |
+
"step": 29
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.09,
|
| 216 |
+
"grad_norm": 0.0887107703024984,
|
| 217 |
+
"learning_rate": 0.000990210535489303,
|
| 218 |
+
"loss": 2.8246,
|
| 219 |
+
"step": 30
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.09,
|
| 223 |
+
"grad_norm": 0.05971709937694414,
|
| 224 |
+
"learning_rate": 0.0009892107390380958,
|
| 225 |
+
"loss": 2.777,
|
| 226 |
+
"step": 31
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.1,
|
| 230 |
+
"grad_norm": 0.04203098841999937,
|
| 231 |
+
"learning_rate": 0.0009881628947712555,
|
| 232 |
+
"loss": 2.7673,
|
| 233 |
+
"step": 32
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.1,
|
| 237 |
+
"grad_norm": 0.03708904853213128,
|
| 238 |
+
"learning_rate": 0.0009870671056027706,
|
| 239 |
+
"loss": 2.7486,
|
| 240 |
+
"step": 33
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.1,
|
| 244 |
+
"grad_norm": 0.03528212659332326,
|
| 245 |
+
"learning_rate": 0.0009859234791555354,
|
| 246 |
+
"loss": 2.7886,
|
| 247 |
+
"step": 34
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.11,
|
| 251 |
+
"grad_norm": 0.03018922862288316,
|
| 252 |
+
"learning_rate": 0.000984732127750782,
|
| 253 |
+
"loss": 2.7327,
|
| 254 |
+
"step": 35
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.11,
|
| 258 |
+
"grad_norm": 0.03686815342528168,
|
| 259 |
+
"learning_rate": 0.0009834931683970467,
|
| 260 |
+
"loss": 2.7474,
|
| 261 |
+
"step": 36
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.11,
|
| 265 |
+
"grad_norm": 0.033172131047386,
|
| 266 |
+
"learning_rate": 0.0009822067227786794,
|
| 267 |
+
"loss": 2.7345,
|
| 268 |
+
"step": 37
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.12,
|
| 272 |
+
"grad_norm": 0.0373016703958135,
|
| 273 |
+
"learning_rate": 0.000980872917243891,
|
| 274 |
+
"loss": 2.7655,
|
| 275 |
+
"step": 38
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.12,
|
| 279 |
+
"grad_norm": 0.03688389598963858,
|
| 280 |
+
"learning_rate": 0.0009794918827923458,
|
| 281 |
+
"loss": 2.7393,
|
| 282 |
+
"step": 39
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.12,
|
| 286 |
+
"grad_norm": 0.038870114526809094,
|
| 287 |
+
"learning_rate": 0.000978063755062294,
|
| 288 |
+
"loss": 2.7611,
|
| 289 |
+
"step": 40
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.13,
|
| 293 |
+
"grad_norm": 0.04038306990198406,
|
| 294 |
+
"learning_rate": 0.0009765886743172511,
|
| 295 |
+
"loss": 2.7386,
|
| 296 |
+
"step": 41
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.13,
|
| 300 |
+
"grad_norm": 0.036670767168349196,
|
| 301 |
+
"learning_rate": 0.0009750667854322206,
|
| 302 |
+
"loss": 2.6912,
|
| 303 |
+
"step": 42
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.13,
|
| 307 |
+
"grad_norm": 0.03291379266422108,
|
| 308 |
+
"learning_rate": 0.0009734982378794661,
|
| 309 |
+
"loss": 2.7719,
|
| 310 |
+
"step": 43
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.13,
|
| 314 |
+
"grad_norm": 0.03542095069122837,
|
| 315 |
+
"learning_rate": 0.0009718831857138308,
|
| 316 |
+
"loss": 2.7095,
|
| 317 |
+
"step": 44
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.14,
|
| 321 |
+
"grad_norm": 0.03556844983258325,
|
| 322 |
+
"learning_rate": 0.0009702217875576068,
|
| 323 |
+
"loss": 2.7435,
|
| 324 |
+
"step": 45
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.14,
|
| 328 |
+
"grad_norm": 0.03705512589107,
|
| 329 |
+
"learning_rate": 0.0009685142065849555,
|
| 330 |
+
"loss": 2.713,
|
| 331 |
+
"step": 46
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.14,
|
| 335 |
+
"grad_norm": 0.027028911527830556,
|
| 336 |
+
"learning_rate": 0.0009667606105058828,
|
| 337 |
+
"loss": 2.6472,
|
| 338 |
+
"step": 47
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.15,
|
| 342 |
+
"grad_norm": 0.028600789964649966,
|
| 343 |
+
"learning_rate": 0.0009649611715497661,
|
| 344 |
+
"loss": 2.6931,
|
| 345 |
+
"step": 48
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.15,
|
| 349 |
+
"grad_norm": 0.029116613327395607,
|
| 350 |
+
"learning_rate": 0.0009631160664484398,
|
| 351 |
+
"loss": 2.7082,
|
| 352 |
+
"step": 49
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.15,
|
| 356 |
+
"grad_norm": 0.02557746875550266,
|
| 357 |
+
"learning_rate": 0.0009612254764188368,
|
| 358 |
+
"loss": 2.6707,
|
| 359 |
+
"step": 50
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.16,
|
| 363 |
+
"grad_norm": 0.0256723041758138,
|
| 364 |
+
"learning_rate": 0.0009592895871451908,
|
| 365 |
+
"loss": 2.6608,
|
| 366 |
+
"step": 51
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.16,
|
| 370 |
+
"grad_norm": 0.02295300029667766,
|
| 371 |
+
"learning_rate": 0.000957308588760799,
|
| 372 |
+
"loss": 2.6942,
|
| 373 |
+
"step": 52
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.16,
|
| 377 |
+
"grad_norm": 0.02203986352323152,
|
| 378 |
+
"learning_rate": 0.0009552826758293487,
|
| 379 |
+
"loss": 2.6441,
|
| 380 |
+
"step": 53
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.16,
|
| 384 |
+
"grad_norm": 0.02299784546740539,
|
| 385 |
+
"learning_rate": 0.0009532120473258075,
|
| 386 |
+
"loss": 2.6728,
|
| 387 |
+
"step": 54
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.17,
|
| 391 |
+
"grad_norm": 0.023913877734993542,
|
| 392 |
+
"learning_rate": 0.0009510969066168813,
|
| 393 |
+
"loss": 2.6924,
|
| 394 |
+
"step": 55
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.17,
|
| 398 |
+
"grad_norm": 0.020131800174103105,
|
| 399 |
+
"learning_rate": 0.0009489374614410414,
|
| 400 |
+
"loss": 2.6151,
|
| 401 |
+
"step": 56
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.17,
|
| 405 |
+
"grad_norm": 0.02106884604663375,
|
| 406 |
+
"learning_rate": 0.0009467339238881198,
|
| 407 |
+
"loss": 2.6413,
|
| 408 |
+
"step": 57
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.18,
|
| 412 |
+
"grad_norm": 0.02440845171243592,
|
| 413 |
+
"learning_rate": 0.0009444865103784803,
|
| 414 |
+
"loss": 2.6663,
|
| 415 |
+
"step": 58
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.18,
|
| 419 |
+
"grad_norm": 0.019024075041881073,
|
| 420 |
+
"learning_rate": 0.0009421954416417624,
|
| 421 |
+
"loss": 2.6063,
|
| 422 |
+
"step": 59
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.18,
|
| 426 |
+
"grad_norm": 0.019667612528268406,
|
| 427 |
+
"learning_rate": 0.0009398609426952018,
|
| 428 |
+
"loss": 2.6697,
|
| 429 |
+
"step": 60
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.19,
|
| 433 |
+
"grad_norm": 0.020324139148677475,
|
| 434 |
+
"learning_rate": 0.0009374832428215309,
|
| 435 |
+
"loss": 2.6739,
|
| 436 |
+
"step": 61
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.19,
|
| 440 |
+
"grad_norm": 0.02049609873784361,
|
| 441 |
+
"learning_rate": 0.00093506257554646,
|
| 442 |
+
"loss": 2.6115,
|
| 443 |
+
"step": 62
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.19,
|
| 447 |
+
"grad_norm": 0.022068793706639576,
|
| 448 |
+
"learning_rate": 0.0009325991786157404,
|
| 449 |
+
"loss": 2.6537,
|
| 450 |
+
"step": 63
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.2,
|
| 454 |
+
"grad_norm": 0.02119215165943166,
|
| 455 |
+
"learning_rate": 0.0009300932939718159,
|
| 456 |
+
"loss": 2.6668,
|
| 457 |
+
"step": 64
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.2,
|
| 461 |
+
"grad_norm": 0.020470851870879974,
|
| 462 |
+
"learning_rate": 0.000927545167730059,
|
| 463 |
+
"loss": 2.6495,
|
| 464 |
+
"step": 65
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.2,
|
| 468 |
+
"grad_norm": 0.020239015102427586,
|
| 469 |
+
"learning_rate": 0.0009249550501545996,
|
| 470 |
+
"loss": 2.6229,
|
| 471 |
+
"step": 66
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.2,
|
| 475 |
+
"grad_norm": 0.02006561036060577,
|
| 476 |
+
"learning_rate": 0.000922323195633745,
|
| 477 |
+
"loss": 2.6541,
|
| 478 |
+
"step": 67
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.21,
|
| 482 |
+
"grad_norm": 0.020517458730459407,
|
| 483 |
+
"learning_rate": 0.0009196498626549943,
|
| 484 |
+
"loss": 2.6263,
|
| 485 |
+
"step": 68
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.21,
|
| 489 |
+
"grad_norm": 0.020546751672942867,
|
| 490 |
+
"learning_rate": 0.0009169353137796533,
|
| 491 |
+
"loss": 2.6868,
|
| 492 |
+
"step": 69
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.21,
|
| 496 |
+
"grad_norm": 0.020795862822058586,
|
| 497 |
+
"learning_rate": 0.0009141798156170446,
|
| 498 |
+
"loss": 2.6376,
|
| 499 |
+
"step": 70
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.22,
|
| 503 |
+
"grad_norm": 0.023603442107788076,
|
| 504 |
+
"learning_rate": 0.0009113836387983239,
|
| 505 |
+
"loss": 2.6679,
|
| 506 |
+
"step": 71
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.22,
|
| 510 |
+
"grad_norm": 0.020593236730693215,
|
| 511 |
+
"learning_rate": 0.0009085470579498995,
|
| 512 |
+
"loss": 2.628,
|
| 513 |
+
"step": 72
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.22,
|
| 517 |
+
"grad_norm": 0.0185506077207348,
|
| 518 |
+
"learning_rate": 0.0009056703516664606,
|
| 519 |
+
"loss": 2.6315,
|
| 520 |
+
"step": 73
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.23,
|
| 524 |
+
"grad_norm": 0.02046637689680032,
|
| 525 |
+
"learning_rate": 0.0009027538024836141,
|
| 526 |
+
"loss": 2.6227,
|
| 527 |
+
"step": 74
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.23,
|
| 531 |
+
"grad_norm": 0.018026412977487194,
|
| 532 |
+
"learning_rate": 0.0008997976968501361,
|
| 533 |
+
"loss": 2.6062,
|
| 534 |
+
"step": 75
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.23,
|
| 538 |
+
"grad_norm": 0.017993838807838707,
|
| 539 |
+
"learning_rate": 0.000896802325099838,
|
| 540 |
+
"loss": 2.5869,
|
| 541 |
+
"step": 76
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.23,
|
| 545 |
+
"grad_norm": 0.01897876413125471,
|
| 546 |
+
"learning_rate": 0.0008937679814230517,
|
| 547 |
+
"loss": 2.6046,
|
| 548 |
+
"step": 77
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.24,
|
| 552 |
+
"grad_norm": 0.019221228785349705,
|
| 553 |
+
"learning_rate": 0.000890694963837735,
|
| 554 |
+
"loss": 2.6163,
|
| 555 |
+
"step": 78
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.24,
|
| 559 |
+
"grad_norm": 0.01999062056670978,
|
| 560 |
+
"learning_rate": 0.0008875835741602029,
|
| 561 |
+
"loss": 2.6624,
|
| 562 |
+
"step": 79
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.24,
|
| 566 |
+
"grad_norm": 0.01688098771768354,
|
| 567 |
+
"learning_rate": 0.0008844341179754839,
|
| 568 |
+
"loss": 2.597,
|
| 569 |
+
"step": 80
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.25,
|
| 573 |
+
"grad_norm": 0.021679649830905732,
|
| 574 |
+
"learning_rate": 0.0008812469046073068,
|
| 575 |
+
"loss": 2.6252,
|
| 576 |
+
"step": 81
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.25,
|
| 580 |
+
"grad_norm": 0.01976703829526433,
|
| 581 |
+
"learning_rate": 0.0008780222470877213,
|
| 582 |
+
"loss": 2.6547,
|
| 583 |
+
"step": 82
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.25,
|
| 587 |
+
"grad_norm": 0.01721216059701167,
|
| 588 |
+
"learning_rate": 0.000874760462126353,
|
| 589 |
+
"loss": 2.5894,
|
| 590 |
+
"step": 83
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.26,
|
| 594 |
+
"grad_norm": 0.017796757505650348,
|
| 595 |
+
"learning_rate": 0.0008714618700792976,
|
| 596 |
+
"loss": 2.6428,
|
| 597 |
+
"step": 84
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.26,
|
| 601 |
+
"grad_norm": 0.017516085193709605,
|
| 602 |
+
"learning_rate": 0.0008681267949176579,
|
| 603 |
+
"loss": 2.6267,
|
| 604 |
+
"step": 85
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.26,
|
| 608 |
+
"grad_norm": 0.019752354059471497,
|
| 609 |
+
"learning_rate": 0.0008647555641957244,
|
| 610 |
+
"loss": 2.6172,
|
| 611 |
+
"step": 86
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.27,
|
| 615 |
+
"grad_norm": 0.01942107114850748,
|
| 616 |
+
"learning_rate": 0.0008613485090188043,
|
| 617 |
+
"loss": 2.6469,
|
| 618 |
+
"step": 87
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.27,
|
| 622 |
+
"grad_norm": 0.021049275004733702,
|
| 623 |
+
"learning_rate": 0.000857905964010703,
|
| 624 |
+
"loss": 2.6337,
|
| 625 |
+
"step": 88
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.27,
|
| 629 |
+
"grad_norm": 0.017375996512513512,
|
| 630 |
+
"learning_rate": 0.0008544282672808579,
|
| 631 |
+
"loss": 2.6066,
|
| 632 |
+
"step": 89
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.27,
|
| 636 |
+
"grad_norm": 0.020054221841685173,
|
| 637 |
+
"learning_rate": 0.0008509157603911319,
|
| 638 |
+
"loss": 2.6293,
|
| 639 |
+
"step": 90
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.28,
|
| 643 |
+
"grad_norm": 0.016754816692027007,
|
| 644 |
+
"learning_rate": 0.0008473687883222664,
|
| 645 |
+
"loss": 2.553,
|
| 646 |
+
"step": 91
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.28,
|
| 650 |
+
"grad_norm": 0.018264517242335677,
|
| 651 |
+
"learning_rate": 0.0008437876994399991,
|
| 652 |
+
"loss": 2.582,
|
| 653 |
+
"step": 92
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.28,
|
| 657 |
+
"grad_norm": 0.01738231752273472,
|
| 658 |
+
"learning_rate": 0.0008401728454608494,
|
| 659 |
+
"loss": 2.6246,
|
| 660 |
+
"step": 93
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.29,
|
| 664 |
+
"grad_norm": 0.017582673986843284,
|
| 665 |
+
"learning_rate": 0.0008365245814175744,
|
| 666 |
+
"loss": 2.5377,
|
| 667 |
+
"step": 94
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.29,
|
| 671 |
+
"grad_norm": 0.01841568278108112,
|
| 672 |
+
"learning_rate": 0.0008328432656242997,
|
| 673 |
+
"loss": 2.6107,
|
| 674 |
+
"step": 95
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.29,
|
| 678 |
+
"grad_norm": 0.01806489585852763,
|
| 679 |
+
"learning_rate": 0.0008291292596413272,
|
| 680 |
+
"loss": 2.5388,
|
| 681 |
+
"step": 96
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.3,
|
| 685 |
+
"grad_norm": 0.01910768409990766,
|
| 686 |
+
"learning_rate": 0.0008253829282396245,
|
| 687 |
+
"loss": 2.6252,
|
| 688 |
+
"step": 97
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.3,
|
| 692 |
+
"grad_norm": 0.019034711943532324,
|
| 693 |
+
"learning_rate": 0.0008216046393649996,
|
| 694 |
+
"loss": 2.6193,
|
| 695 |
+
"step": 98
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.3,
|
| 699 |
+
"grad_norm": 0.020284737834854932,
|
| 700 |
+
"learning_rate": 0.0008177947641019621,
|
| 701 |
+
"loss": 2.6149,
|
| 702 |
+
"step": 99
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.3,
|
| 706 |
+
"grad_norm": 0.05158744985589759,
|
| 707 |
+
"learning_rate": 0.0008139536766372775,
|
| 708 |
+
"loss": 2.6072,
|
| 709 |
+
"step": 100
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.31,
|
| 713 |
+
"grad_norm": 0.018922120101005344,
|
| 714 |
+
"learning_rate": 0.0008100817542232173,
|
| 715 |
+
"loss": 2.6032,
|
| 716 |
+
"step": 101
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.31,
|
| 720 |
+
"grad_norm": 0.019044369144979888,
|
| 721 |
+
"learning_rate": 0.000806179377140506,
|
| 722 |
+
"loss": 2.5879,
|
| 723 |
+
"step": 102
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.31,
|
| 727 |
+
"grad_norm": 0.017582902915047,
|
| 728 |
+
"learning_rate": 0.000802246928660972,
|
| 729 |
+
"loss": 2.6069,
|
| 730 |
+
"step": 103
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.32,
|
| 734 |
+
"grad_norm": 0.019578808618249007,
|
| 735 |
+
"learning_rate": 0.0007982847950099055,
|
| 736 |
+
"loss": 2.5965,
|
| 737 |
+
"step": 104
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.32,
|
| 741 |
+
"grad_norm": 0.018236739907291074,
|
| 742 |
+
"learning_rate": 0.0007942933653281245,
|
| 743 |
+
"loss": 2.5707,
|
| 744 |
+
"step": 105
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.32,
|
| 748 |
+
"grad_norm": 0.018258178535900983,
|
| 749 |
+
"learning_rate": 0.0007902730316337556,
|
| 750 |
+
"loss": 2.5382,
|
| 751 |
+
"step": 106
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.33,
|
| 755 |
+
"grad_norm": 0.022391092471974905,
|
| 756 |
+
"learning_rate": 0.0007862241887837322,
|
| 757 |
+
"loss": 2.6118,
|
| 758 |
+
"step": 107
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.33,
|
| 762 |
+
"grad_norm": 0.016821289155497087,
|
| 763 |
+
"learning_rate": 0.0007821472344350131,
|
| 764 |
+
"loss": 2.5829,
|
| 765 |
+
"step": 108
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.33,
|
| 769 |
+
"grad_norm": 0.019386550472160386,
|
| 770 |
+
"learning_rate": 0.0007780425690055274,
|
| 771 |
+
"loss": 2.6255,
|
| 772 |
+
"step": 109
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.34,
|
| 776 |
+
"grad_norm": 0.1772374509222317,
|
| 777 |
+
"learning_rate": 0.0007739105956348464,
|
| 778 |
+
"loss": 2.5617,
|
| 779 |
+
"step": 110
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.34,
|
| 783 |
+
"grad_norm": 0.020123188408530047,
|
| 784 |
+
"learning_rate": 0.0007697517201445905,
|
| 785 |
+
"loss": 2.6127,
|
| 786 |
+
"step": 111
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.34,
|
| 790 |
+
"grad_norm": 0.0217268258079,
|
| 791 |
+
"learning_rate": 0.0007655663509985707,
|
| 792 |
+
"loss": 2.5991,
|
| 793 |
+
"step": 112
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.34,
|
| 797 |
+
"grad_norm": 0.018981113976254002,
|
| 798 |
+
"learning_rate": 0.0007613548992626711,
|
| 799 |
+
"loss": 2.6047,
|
| 800 |
+
"step": 113
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.35,
|
| 804 |
+
"grad_norm": 0.01878803361275613,
|
| 805 |
+
"learning_rate": 0.0007571177785644766,
|
| 806 |
+
"loss": 2.5482,
|
| 807 |
+
"step": 114
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.35,
|
| 811 |
+
"grad_norm": 0.018908774218863146,
|
| 812 |
+
"learning_rate": 0.0007528554050526488,
|
| 813 |
+
"loss": 2.6141,
|
| 814 |
+
"step": 115
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.35,
|
| 818 |
+
"grad_norm": 0.019547294966543106,
|
| 819 |
+
"learning_rate": 0.0007485681973560532,
|
| 820 |
+
"loss": 2.5407,
|
| 821 |
+
"step": 116
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.36,
|
| 825 |
+
"grad_norm": 0.018739519276120584,
|
| 826 |
+
"learning_rate": 0.0007442565765426436,
|
| 827 |
+
"loss": 2.6212,
|
| 828 |
+
"step": 117
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.36,
|
| 832 |
+
"grad_norm": 0.018979109952080055,
|
| 833 |
+
"learning_rate": 0.0007399209660781074,
|
| 834 |
+
"loss": 2.5721,
|
| 835 |
+
"step": 118
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.36,
|
| 839 |
+
"grad_norm": 0.0175358024051344,
|
| 840 |
+
"learning_rate": 0.0007355617917842751,
|
| 841 |
+
"loss": 2.577,
|
| 842 |
+
"step": 119
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.37,
|
| 846 |
+
"grad_norm": 0.018692342003597935,
|
| 847 |
+
"learning_rate": 0.0007311794817972975,
|
| 848 |
+
"loss": 2.5944,
|
| 849 |
+
"step": 120
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.37,
|
| 853 |
+
"grad_norm": 0.019650294357122313,
|
| 854 |
+
"learning_rate": 0.0007267744665255965,
|
| 855 |
+
"loss": 2.608,
|
| 856 |
+
"step": 121
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.37,
|
| 860 |
+
"grad_norm": 0.01787007786983172,
|
| 861 |
+
"learning_rate": 0.0007223471786075934,
|
| 862 |
+
"loss": 2.5898,
|
| 863 |
+
"step": 122
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.38,
|
| 867 |
+
"grad_norm": 0.018327854729029175,
|
| 868 |
+
"learning_rate": 0.0007178980528692161,
|
| 869 |
+
"loss": 2.5641,
|
| 870 |
+
"step": 123
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.38,
|
| 874 |
+
"grad_norm": 0.01819571331500734,
|
| 875 |
+
"learning_rate": 0.0007134275262811934,
|
| 876 |
+
"loss": 2.5724,
|
| 877 |
+
"step": 124
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.38,
|
| 881 |
+
"grad_norm": 0.016879527517399177,
|
| 882 |
+
"learning_rate": 0.0007089360379161381,
|
| 883 |
+
"loss": 2.6016,
|
| 884 |
+
"step": 125
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.38,
|
| 888 |
+
"grad_norm": 0.018893461615821387,
|
| 889 |
+
"learning_rate": 0.0007044240289054227,
|
| 890 |
+
"loss": 2.5726,
|
| 891 |
+
"step": 126
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.39,
|
| 895 |
+
"grad_norm": 0.018652185039036564,
|
| 896 |
+
"learning_rate": 0.0006998919423958547,
|
| 897 |
+
"loss": 2.5559,
|
| 898 |
+
"step": 127
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.39,
|
| 902 |
+
"grad_norm": 0.020673958617829544,
|
| 903 |
+
"learning_rate": 0.0006953402235061519,
|
| 904 |
+
"loss": 2.634,
|
| 905 |
+
"step": 128
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.39,
|
| 909 |
+
"grad_norm": 0.01939109964021446,
|
| 910 |
+
"learning_rate": 0.0006907693192832263,
|
| 911 |
+
"loss": 2.568,
|
| 912 |
+
"step": 129
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.4,
|
| 916 |
+
"grad_norm": 0.018062900613076246,
|
| 917 |
+
"learning_rate": 0.0006861796786582761,
|
| 918 |
+
"loss": 2.554,
|
| 919 |
+
"step": 130
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.4,
|
| 923 |
+
"grad_norm": 0.018609687828424634,
|
| 924 |
+
"learning_rate": 0.0006815717524026949,
|
| 925 |
+
"loss": 2.5842,
|
| 926 |
+
"step": 131
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.4,
|
| 930 |
+
"grad_norm": 0.019554131820221306,
|
| 931 |
+
"learning_rate": 0.0006769459930837989,
|
| 932 |
+
"loss": 2.632,
|
| 933 |
+
"step": 132
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.41,
|
| 937 |
+
"grad_norm": 0.017713236616086883,
|
| 938 |
+
"learning_rate": 0.0006723028550203778,
|
| 939 |
+
"loss": 2.5547,
|
| 940 |
+
"step": 133
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.41,
|
| 944 |
+
"grad_norm": 0.019353124709199913,
|
| 945 |
+
"learning_rate": 0.000667642794238074,
|
| 946 |
+
"loss": 2.6321,
|
| 947 |
+
"step": 134
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.41,
|
| 951 |
+
"grad_norm": 0.01984763650797756,
|
| 952 |
+
"learning_rate": 0.0006629662684245948,
|
| 953 |
+
"loss": 2.5521,
|
| 954 |
+
"step": 135
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.41,
|
| 958 |
+
"grad_norm": 0.0194768410660429,
|
| 959 |
+
"learning_rate": 0.0006582737368847592,
|
| 960 |
+
"loss": 2.564,
|
| 961 |
+
"step": 136
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.42,
|
| 965 |
+
"grad_norm": 0.02008541514106773,
|
| 966 |
+
"learning_rate": 0.0006535656604953884,
|
| 967 |
+
"loss": 2.6406,
|
| 968 |
+
"step": 137
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.42,
|
| 972 |
+
"grad_norm": 0.019023120632534838,
|
| 973 |
+
"learning_rate": 0.0006488425016600402,
|
| 974 |
+
"loss": 2.5492,
|
| 975 |
+
"step": 138
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.42,
|
| 979 |
+
"grad_norm": 0.019894786782134775,
|
| 980 |
+
"learning_rate": 0.0006441047242635947,
|
| 981 |
+
"loss": 2.598,
|
| 982 |
+
"step": 139
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.43,
|
| 986 |
+
"grad_norm": 0.02011873025218978,
|
| 987 |
+
"learning_rate": 0.0006393527936266933,
|
| 988 |
+
"loss": 2.5729,
|
| 989 |
+
"step": 140
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.43,
|
| 993 |
+
"grad_norm": 0.019172173916513022,
|
| 994 |
+
"learning_rate": 0.0006345871764600374,
|
| 995 |
+
"loss": 2.5594,
|
| 996 |
+
"step": 141
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.43,
|
| 1000 |
+
"grad_norm": 0.019721366832981004,
|
| 1001 |
+
"learning_rate": 0.0006298083408185502,
|
| 1002 |
+
"loss": 2.5915,
|
| 1003 |
+
"step": 142
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.44,
|
| 1007 |
+
"grad_norm": 0.020228287976845075,
|
| 1008 |
+
"learning_rate": 0.0006250167560554076,
|
| 1009 |
+
"loss": 2.5898,
|
| 1010 |
+
"step": 143
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.44,
|
| 1014 |
+
"grad_norm": 0.020882340174208855,
|
| 1015 |
+
"learning_rate": 0.0006202128927759391,
|
| 1016 |
+
"loss": 2.6273,
|
| 1017 |
+
"step": 144
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.44,
|
| 1021 |
+
"grad_norm": 0.02081451600660678,
|
| 1022 |
+
"learning_rate": 0.0006153972227914089,
|
| 1023 |
+
"loss": 2.595,
|
| 1024 |
+
"step": 145
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.45,
|
| 1028 |
+
"grad_norm": 0.01941546346347613,
|
| 1029 |
+
"learning_rate": 0.0006105702190726764,
|
| 1030 |
+
"loss": 2.5732,
|
| 1031 |
+
"step": 146
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.45,
|
| 1035 |
+
"grad_norm": 0.02166634263027907,
|
| 1036 |
+
"learning_rate": 0.000605732355703743,
|
| 1037 |
+
"loss": 2.5453,
|
| 1038 |
+
"step": 147
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.45,
|
| 1042 |
+
"grad_norm": 0.019573148109730217,
|
| 1043 |
+
"learning_rate": 0.0006008841078351903,
|
| 1044 |
+
"loss": 2.5429,
|
| 1045 |
+
"step": 148
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.45,
|
| 1049 |
+
"grad_norm": 0.01903540723775177,
|
| 1050 |
+
"learning_rate": 0.0005960259516375134,
|
| 1051 |
+
"loss": 2.5388,
|
| 1052 |
+
"step": 149
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.46,
|
| 1056 |
+
"grad_norm": 0.02051218709860799,
|
| 1057 |
+
"learning_rate": 0.0005911583642543531,
|
| 1058 |
+
"loss": 2.5793,
|
| 1059 |
+
"step": 150
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.46,
|
| 1063 |
+
"grad_norm": 0.02101151521755854,
|
| 1064 |
+
"learning_rate": 0.0005862818237556344,
|
| 1065 |
+
"loss": 2.5677,
|
| 1066 |
+
"step": 151
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.46,
|
| 1070 |
+
"grad_norm": 0.02162618990843192,
|
| 1071 |
+
"learning_rate": 0.0005813968090906116,
|
| 1072 |
+
"loss": 2.5635,
|
| 1073 |
+
"step": 152
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.47,
|
| 1077 |
+
"grad_norm": 0.018935934203135756,
|
| 1078 |
+
"learning_rate": 0.0005765038000408295,
|
| 1079 |
+
"loss": 2.5174,
|
| 1080 |
+
"step": 153
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.47,
|
| 1084 |
+
"grad_norm": 0.021177497387846266,
|
| 1085 |
+
"learning_rate": 0.0005716032771730008,
|
| 1086 |
+
"loss": 2.5266,
|
| 1087 |
+
"step": 154
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.47,
|
| 1091 |
+
"grad_norm": 0.019633074877257486,
|
| 1092 |
+
"learning_rate": 0.0005666957217918076,
|
| 1093 |
+
"loss": 2.5909,
|
| 1094 |
+
"step": 155
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.48,
|
| 1098 |
+
"grad_norm": 0.020321617227441156,
|
| 1099 |
+
"learning_rate": 0.0005617816158926302,
|
| 1100 |
+
"loss": 2.5727,
|
| 1101 |
+
"step": 156
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.48,
|
| 1105 |
+
"grad_norm": 0.02171314000863455,
|
| 1106 |
+
"learning_rate": 0.0005568614421142077,
|
| 1107 |
+
"loss": 2.5728,
|
| 1108 |
+
"step": 157
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.48,
|
| 1112 |
+
"grad_norm": 0.02036197464612551,
|
| 1113 |
+
"learning_rate": 0.0005519356836912357,
|
| 1114 |
+
"loss": 2.5546,
|
| 1115 |
+
"step": 158
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.48,
|
| 1119 |
+
"grad_norm": 0.021214591076630044,
|
| 1120 |
+
"learning_rate": 0.0005470048244069055,
|
| 1121 |
+
"loss": 2.6089,
|
| 1122 |
+
"step": 159
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.49,
|
| 1126 |
+
"grad_norm": 0.021380356809754895,
|
| 1127 |
+
"learning_rate": 0.0005420693485453892,
|
| 1128 |
+
"loss": 2.5701,
|
| 1129 |
+
"step": 160
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.49,
|
| 1133 |
+
"grad_norm": 0.020301144411914027,
|
| 1134 |
+
"learning_rate": 0.0005371297408442765,
|
| 1135 |
+
"loss": 2.5783,
|
| 1136 |
+
"step": 161
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.49,
|
| 1140 |
+
"grad_norm": 0.02317064813036681,
|
| 1141 |
+
"learning_rate": 0.0005321864864469646,
|
| 1142 |
+
"loss": 2.5787,
|
| 1143 |
+
"step": 162
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.5,
|
| 1147 |
+
"grad_norm": 0.020981576388978553,
|
| 1148 |
+
"learning_rate": 0.0005272400708550113,
|
| 1149 |
+
"loss": 2.5657,
|
| 1150 |
+
"step": 163
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.5,
|
| 1154 |
+
"grad_norm": 0.01972062092753295,
|
| 1155 |
+
"learning_rate": 0.0005222909798804515,
|
| 1156 |
+
"loss": 2.5941,
|
| 1157 |
+
"step": 164
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.5,
|
| 1161 |
+
"grad_norm": 0.022818553102453083,
|
| 1162 |
+
"learning_rate": 0.0005173396995980818,
|
| 1163 |
+
"loss": 2.5567,
|
| 1164 |
+
"step": 165
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.51,
|
| 1168 |
+
"grad_norm": 0.021333416992185616,
|
| 1169 |
+
"learning_rate": 0.0005123867162977224,
|
| 1170 |
+
"loss": 2.5552,
|
| 1171 |
+
"step": 166
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.51,
|
| 1175 |
+
"grad_norm": 0.020142143404330078,
|
| 1176 |
+
"learning_rate": 0.0005074325164364548,
|
| 1177 |
+
"loss": 2.5305,
|
| 1178 |
+
"step": 167
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.51,
|
| 1182 |
+
"grad_norm": 0.022025222387284903,
|
| 1183 |
+
"learning_rate": 0.0005024775865908451,
|
| 1184 |
+
"loss": 2.59,
|
| 1185 |
+
"step": 168
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.52,
|
| 1189 |
+
"grad_norm": 0.020907939242901075,
|
| 1190 |
+
"learning_rate": 0.000497522413409155,
|
| 1191 |
+
"loss": 2.5452,
|
| 1192 |
+
"step": 169
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.52,
|
| 1196 |
+
"grad_norm": 0.022927802322477854,
|
| 1197 |
+
"learning_rate": 0.0004925674835635454,
|
| 1198 |
+
"loss": 2.5792,
|
| 1199 |
+
"step": 170
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.52,
|
| 1203 |
+
"grad_norm": 0.020657968617921074,
|
| 1204 |
+
"learning_rate": 0.00048761328370227773,
|
| 1205 |
+
"loss": 2.5099,
|
| 1206 |
+
"step": 171
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.52,
|
| 1210 |
+
"grad_norm": 0.020110997895501886,
|
| 1211 |
+
"learning_rate": 0.0004826603004019182,
|
| 1212 |
+
"loss": 2.5163,
|
| 1213 |
+
"step": 172
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.53,
|
| 1217 |
+
"grad_norm": 0.021049174068166514,
|
| 1218 |
+
"learning_rate": 0.0004777090201195486,
|
| 1219 |
+
"loss": 2.5372,
|
| 1220 |
+
"step": 173
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.53,
|
| 1224 |
+
"grad_norm": 0.02117182813897076,
|
| 1225 |
+
"learning_rate": 0.00047275992914498865,
|
| 1226 |
+
"loss": 2.5022,
|
| 1227 |
+
"step": 174
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.53,
|
| 1231 |
+
"grad_norm": 0.020475434729428973,
|
| 1232 |
+
"learning_rate": 0.0004678135135530355,
|
| 1233 |
+
"loss": 2.5626,
|
| 1234 |
+
"step": 175
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.54,
|
| 1238 |
+
"grad_norm": 0.01948270633034859,
|
| 1239 |
+
"learning_rate": 0.0004628702591557237,
|
| 1240 |
+
"loss": 2.5286,
|
| 1241 |
+
"step": 176
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.54,
|
| 1245 |
+
"grad_norm": 0.020212144683261414,
|
| 1246 |
+
"learning_rate": 0.00045793065145461064,
|
| 1247 |
+
"loss": 2.5305,
|
| 1248 |
+
"step": 177
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 0.54,
|
| 1252 |
+
"grad_norm": 0.020337213413087113,
|
| 1253 |
+
"learning_rate": 0.00045299517559309457,
|
| 1254 |
+
"loss": 2.5624,
|
| 1255 |
+
"step": 178
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.55,
|
| 1259 |
+
"grad_norm": 0.02101629597235714,
|
| 1260 |
+
"learning_rate": 0.00044806431630876436,
|
| 1261 |
+
"loss": 2.5109,
|
| 1262 |
+
"step": 179
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.55,
|
| 1266 |
+
"grad_norm": 0.02099914834706713,
|
| 1267 |
+
"learning_rate": 0.00044313855788579234,
|
| 1268 |
+
"loss": 2.5702,
|
| 1269 |
+
"step": 180
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.55,
|
| 1273 |
+
"grad_norm": 0.020635070288327804,
|
| 1274 |
+
"learning_rate": 0.0004382183841073698,
|
| 1275 |
+
"loss": 2.5483,
|
| 1276 |
+
"step": 181
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.56,
|
| 1280 |
+
"grad_norm": 0.019941103414359743,
|
| 1281 |
+
"learning_rate": 0.00043330427820819256,
|
| 1282 |
+
"loss": 2.5722,
|
| 1283 |
+
"step": 182
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.56,
|
| 1287 |
+
"grad_norm": 0.032674916215766084,
|
| 1288 |
+
"learning_rate": 0.0004283967228269992,
|
| 1289 |
+
"loss": 2.5775,
|
| 1290 |
+
"step": 183
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.56,
|
| 1294 |
+
"grad_norm": 0.021496384704403824,
|
| 1295 |
+
"learning_rate": 0.00042349619995917057,
|
| 1296 |
+
"loss": 2.5856,
|
| 1297 |
+
"step": 184
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 0.56,
|
| 1301 |
+
"grad_norm": 0.030321207463772364,
|
| 1302 |
+
"learning_rate": 0.0004186031909093884,
|
| 1303 |
+
"loss": 2.5618,
|
| 1304 |
+
"step": 185
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.57,
|
| 1308 |
+
"grad_norm": 0.02175980169134122,
|
| 1309 |
+
"learning_rate": 0.00041371817624436577,
|
| 1310 |
+
"loss": 2.5437,
|
| 1311 |
+
"step": 186
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.57,
|
| 1315 |
+
"grad_norm": 0.020391275857096873,
|
| 1316 |
+
"learning_rate": 0.0004088416357456471,
|
| 1317 |
+
"loss": 2.5338,
|
| 1318 |
+
"step": 187
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 0.57,
|
| 1322 |
+
"grad_norm": 0.022437463329837284,
|
| 1323 |
+
"learning_rate": 0.00040397404836248684,
|
| 1324 |
+
"loss": 2.5812,
|
| 1325 |
+
"step": 188
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.58,
|
| 1329 |
+
"grad_norm": 0.02129701651167051,
|
| 1330 |
+
"learning_rate": 0.0003991158921648096,
|
| 1331 |
+
"loss": 2.5453,
|
| 1332 |
+
"step": 189
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 0.58,
|
| 1336 |
+
"grad_norm": 0.0238633278616985,
|
| 1337 |
+
"learning_rate": 0.00039426764429625693,
|
| 1338 |
+
"loss": 2.5233,
|
| 1339 |
+
"step": 190
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 0.58,
|
| 1343 |
+
"grad_norm": 0.02998116183042863,
|
| 1344 |
+
"learning_rate": 0.0003894297809273237,
|
| 1345 |
+
"loss": 2.5646,
|
| 1346 |
+
"step": 191
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.59,
|
| 1350 |
+
"grad_norm": 0.022238039030238917,
|
| 1351 |
+
"learning_rate": 0.00038460277720859116,
|
| 1352 |
+
"loss": 2.5242,
|
| 1353 |
+
"step": 192
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.59,
|
| 1357 |
+
"grad_norm": 0.0231977730809198,
|
| 1358 |
+
"learning_rate": 0.00037978710722406115,
|
| 1359 |
+
"loss": 2.5681,
|
| 1360 |
+
"step": 193
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.59,
|
| 1364 |
+
"grad_norm": 1.3831873669660701,
|
| 1365 |
+
"learning_rate": 0.0003749832439445925,
|
| 1366 |
+
"loss": 2.6152,
|
| 1367 |
+
"step": 194
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.59,
|
| 1371 |
+
"grad_norm": 0.024837482288635006,
|
| 1372 |
+
"learning_rate": 0.0003701916591814497,
|
| 1373 |
+
"loss": 2.5305,
|
| 1374 |
+
"step": 195
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.6,
|
| 1378 |
+
"grad_norm": 0.02460819770267967,
|
| 1379 |
+
"learning_rate": 0.00036541282353996275,
|
| 1380 |
+
"loss": 2.5512,
|
| 1381 |
+
"step": 196
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 0.6,
|
| 1385 |
+
"grad_norm": 0.021382499301954485,
|
| 1386 |
+
"learning_rate": 0.0003606472063733067,
|
| 1387 |
+
"loss": 2.5726,
|
| 1388 |
+
"step": 197
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.6,
|
| 1392 |
+
"grad_norm": 0.022843480204416112,
|
| 1393 |
+
"learning_rate": 0.00035589527573640534,
|
| 1394 |
+
"loss": 2.5911,
|
| 1395 |
+
"step": 198
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.61,
|
| 1399 |
+
"grad_norm": 0.0217433912155617,
|
| 1400 |
+
"learning_rate": 0.0003511574983399599,
|
| 1401 |
+
"loss": 2.5437,
|
| 1402 |
+
"step": 199
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.61,
|
| 1406 |
+
"grad_norm": 0.022341313083280037,
|
| 1407 |
+
"learning_rate": 0.0003464343395046117,
|
| 1408 |
+
"loss": 2.5565,
|
| 1409 |
+
"step": 200
|
| 1410 |
+
}
|
| 1411 |
+
],
|
| 1412 |
+
"logging_steps": 1.0,
|
| 1413 |
+
"max_steps": 327,
|
| 1414 |
+
"num_input_tokens_seen": 0,
|
| 1415 |
+
"num_train_epochs": 1,
|
| 1416 |
+
"save_steps": 25,
|
| 1417 |
+
"total_flos": 3.540319303825883e+18,
|
| 1418 |
+
"train_batch_size": 16,
|
| 1419 |
+
"trial_name": null,
|
| 1420 |
+
"trial_params": null
|
| 1421 |
+
}
|