Upload folder using huggingface_hub
Browse files- config.json +43 -0
- mm_projector.bin +3 -0
- trainer_state.json +2620 -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": "raw_bert_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/clip-patch14-224",
|
| 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": -2,
|
| 27 |
+
"mm_vision_tower": "/model_zoo/openai-clip-vit-large-patch14",
|
| 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": true,
|
| 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 sha256:0dae2753415fac2ef04876cf59cc4d49ed2fef63b1ddeea8335613cbf063c05f
|
| 3 |
+
size 381270945
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2620 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9993247805536799,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 370,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"grad_norm": 21.415120223600994,
|
| 14 |
+
"learning_rate": 4.1666666666666665e-05,
|
| 15 |
+
"loss": 8.6562,
|
| 16 |
+
"step": 1
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.01,
|
| 20 |
+
"grad_norm": 13.773433414832953,
|
| 21 |
+
"learning_rate": 8.333333333333333e-05,
|
| 22 |
+
"loss": 8.7148,
|
| 23 |
+
"step": 2
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.01,
|
| 27 |
+
"grad_norm": 5.8226925278473605,
|
| 28 |
+
"learning_rate": 0.000125,
|
| 29 |
+
"loss": 6.1523,
|
| 30 |
+
"step": 3
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.01,
|
| 34 |
+
"grad_norm": 6.913578305289376,
|
| 35 |
+
"learning_rate": 0.00016666666666666666,
|
| 36 |
+
"loss": 4.8105,
|
| 37 |
+
"step": 4
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.01,
|
| 41 |
+
"grad_norm": 7.564457947295859,
|
| 42 |
+
"learning_rate": 0.00020833333333333335,
|
| 43 |
+
"loss": 4.7148,
|
| 44 |
+
"step": 5
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.02,
|
| 48 |
+
"grad_norm": 3.4572492282918157,
|
| 49 |
+
"learning_rate": 0.00025,
|
| 50 |
+
"loss": 4.3516,
|
| 51 |
+
"step": 6
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.02,
|
| 55 |
+
"grad_norm": 2.1297987720642504,
|
| 56 |
+
"learning_rate": 0.0002916666666666667,
|
| 57 |
+
"loss": 3.9824,
|
| 58 |
+
"step": 7
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.02,
|
| 62 |
+
"grad_norm": 1.1136160564583724,
|
| 63 |
+
"learning_rate": 0.0003333333333333333,
|
| 64 |
+
"loss": 3.7812,
|
| 65 |
+
"step": 8
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.02,
|
| 69 |
+
"grad_norm": 0.8904819206285229,
|
| 70 |
+
"learning_rate": 0.000375,
|
| 71 |
+
"loss": 3.7344,
|
| 72 |
+
"step": 9
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.03,
|
| 76 |
+
"grad_norm": 2.11038283184139,
|
| 77 |
+
"learning_rate": 0.0004166666666666667,
|
| 78 |
+
"loss": 3.7041,
|
| 79 |
+
"step": 10
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.03,
|
| 83 |
+
"grad_norm": 0.802508097218576,
|
| 84 |
+
"learning_rate": 0.0004583333333333333,
|
| 85 |
+
"loss": 3.6377,
|
| 86 |
+
"step": 11
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.03,
|
| 90 |
+
"grad_norm": 0.4105103213318609,
|
| 91 |
+
"learning_rate": 0.0005,
|
| 92 |
+
"loss": 3.5898,
|
| 93 |
+
"step": 12
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.04,
|
| 97 |
+
"grad_norm": 0.28889375510308085,
|
| 98 |
+
"learning_rate": 0.0004999903741094759,
|
| 99 |
+
"loss": 3.5059,
|
| 100 |
+
"step": 13
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.04,
|
| 104 |
+
"grad_norm": 0.1903491115661804,
|
| 105 |
+
"learning_rate": 0.0004999614971791658,
|
| 106 |
+
"loss": 3.5312,
|
| 107 |
+
"step": 14
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.04,
|
| 111 |
+
"grad_norm": 0.18406971686632234,
|
| 112 |
+
"learning_rate": 0.0004999133714327992,
|
| 113 |
+
"loss": 3.5264,
|
| 114 |
+
"step": 15
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.04,
|
| 118 |
+
"grad_norm": 0.19187377072404146,
|
| 119 |
+
"learning_rate": 0.0004998460005764011,
|
| 120 |
+
"loss": 3.4863,
|
| 121 |
+
"step": 16
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.05,
|
| 125 |
+
"grad_norm": 0.18894966967720278,
|
| 126 |
+
"learning_rate": 0.0004997593897980076,
|
| 127 |
+
"loss": 3.5098,
|
| 128 |
+
"step": 17
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.05,
|
| 132 |
+
"grad_norm": 0.16990770049278048,
|
| 133 |
+
"learning_rate": 0.0004996535457672657,
|
| 134 |
+
"loss": 3.46,
|
| 135 |
+
"step": 18
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.05,
|
| 139 |
+
"grad_norm": 0.14341359539921433,
|
| 140 |
+
"learning_rate": 0.0004995284766349196,
|
| 141 |
+
"loss": 3.4102,
|
| 142 |
+
"step": 19
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.05,
|
| 146 |
+
"grad_norm": 0.13853022483240476,
|
| 147 |
+
"learning_rate": 0.0004993841920321838,
|
| 148 |
+
"loss": 3.3721,
|
| 149 |
+
"step": 20
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.06,
|
| 153 |
+
"grad_norm": 0.14882163330247777,
|
| 154 |
+
"learning_rate": 0.0004992207030700005,
|
| 155 |
+
"loss": 3.4062,
|
| 156 |
+
"step": 21
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.06,
|
| 160 |
+
"grad_norm": 0.12082174803609903,
|
| 161 |
+
"learning_rate": 0.0004990380223381846,
|
| 162 |
+
"loss": 3.3682,
|
| 163 |
+
"step": 22
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.06,
|
| 167 |
+
"grad_norm": 0.11072980253911602,
|
| 168 |
+
"learning_rate": 0.0004988361639044537,
|
| 169 |
+
"loss": 3.335,
|
| 170 |
+
"step": 23
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.06,
|
| 174 |
+
"grad_norm": 0.10689352906545974,
|
| 175 |
+
"learning_rate": 0.0004986151433133455,
|
| 176 |
+
"loss": 3.3447,
|
| 177 |
+
"step": 24
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.07,
|
| 181 |
+
"grad_norm": 0.10490544779612036,
|
| 182 |
+
"learning_rate": 0.0004983749775850199,
|
| 183 |
+
"loss": 3.3252,
|
| 184 |
+
"step": 25
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.07,
|
| 188 |
+
"grad_norm": 0.10182457932796082,
|
| 189 |
+
"learning_rate": 0.0004981156852139491,
|
| 190 |
+
"loss": 3.3037,
|
| 191 |
+
"step": 26
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.07,
|
| 195 |
+
"grad_norm": 0.1056648298216298,
|
| 196 |
+
"learning_rate": 0.0004978372861674928,
|
| 197 |
+
"loss": 3.3164,
|
| 198 |
+
"step": 27
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.08,
|
| 202 |
+
"grad_norm": 0.08205846396464361,
|
| 203 |
+
"learning_rate": 0.0004975398018843612,
|
| 204 |
+
"loss": 3.2939,
|
| 205 |
+
"step": 28
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.08,
|
| 209 |
+
"grad_norm": 0.08913381428371997,
|
| 210 |
+
"learning_rate": 0.0004972232552729632,
|
| 211 |
+
"loss": 3.2949,
|
| 212 |
+
"step": 29
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.08,
|
| 216 |
+
"grad_norm": 0.07539237478136081,
|
| 217 |
+
"learning_rate": 0.000496887670709643,
|
| 218 |
+
"loss": 3.2461,
|
| 219 |
+
"step": 30
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.08,
|
| 223 |
+
"grad_norm": 0.07628363158131596,
|
| 224 |
+
"learning_rate": 0.0004965330740368029,
|
| 225 |
+
"loss": 3.2578,
|
| 226 |
+
"step": 31
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.09,
|
| 230 |
+
"grad_norm": 0.07982185655162036,
|
| 231 |
+
"learning_rate": 0.0004961594925609129,
|
| 232 |
+
"loss": 3.2451,
|
| 233 |
+
"step": 32
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.09,
|
| 237 |
+
"grad_norm": 0.08695759890018029,
|
| 238 |
+
"learning_rate": 0.000495766955050408,
|
| 239 |
+
"loss": 3.2607,
|
| 240 |
+
"step": 33
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.09,
|
| 244 |
+
"grad_norm": 0.06301153004824052,
|
| 245 |
+
"learning_rate": 0.0004953554917334731,
|
| 246 |
+
"loss": 3.2061,
|
| 247 |
+
"step": 34
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.09,
|
| 251 |
+
"grad_norm": 0.06474729877299198,
|
| 252 |
+
"learning_rate": 0.0004949251342957151,
|
| 253 |
+
"loss": 3.25,
|
| 254 |
+
"step": 35
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.1,
|
| 258 |
+
"grad_norm": 0.07883442367683367,
|
| 259 |
+
"learning_rate": 0.0004944759158777223,
|
| 260 |
+
"loss": 3.2236,
|
| 261 |
+
"step": 36
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.1,
|
| 265 |
+
"grad_norm": 0.07015055529327788,
|
| 266 |
+
"learning_rate": 0.0004940078710725135,
|
| 267 |
+
"loss": 3.2227,
|
| 268 |
+
"step": 37
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.1,
|
| 272 |
+
"grad_norm": 0.07119572625376963,
|
| 273 |
+
"learning_rate": 0.0004935210359228729,
|
| 274 |
+
"loss": 3.2031,
|
| 275 |
+
"step": 38
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.11,
|
| 279 |
+
"grad_norm": 0.06367666239627473,
|
| 280 |
+
"learning_rate": 0.0004930154479185756,
|
| 281 |
+
"loss": 3.252,
|
| 282 |
+
"step": 39
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.11,
|
| 286 |
+
"grad_norm": 0.06300151867440586,
|
| 287 |
+
"learning_rate": 0.0004924911459934997,
|
| 288 |
+
"loss": 3.1777,
|
| 289 |
+
"step": 40
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.11,
|
| 293 |
+
"grad_norm": 0.07472850979326924,
|
| 294 |
+
"learning_rate": 0.0004919481705226285,
|
| 295 |
+
"loss": 3.2314,
|
| 296 |
+
"step": 41
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.11,
|
| 300 |
+
"grad_norm": 0.08114730974850456,
|
| 301 |
+
"learning_rate": 0.0004913865633189419,
|
| 302 |
+
"loss": 3.2119,
|
| 303 |
+
"step": 42
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.12,
|
| 307 |
+
"grad_norm": 0.05895132991107217,
|
| 308 |
+
"learning_rate": 0.0004908063676301953,
|
| 309 |
+
"loss": 3.2188,
|
| 310 |
+
"step": 43
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.12,
|
| 314 |
+
"grad_norm": 0.05010637153257128,
|
| 315 |
+
"learning_rate": 0.0004902076281355902,
|
| 316 |
+
"loss": 3.1689,
|
| 317 |
+
"step": 44
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.12,
|
| 321 |
+
"grad_norm": 0.06316741853726789,
|
| 322 |
+
"learning_rate": 0.0004895903909423332,
|
| 323 |
+
"loss": 3.1904,
|
| 324 |
+
"step": 45
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.12,
|
| 328 |
+
"grad_norm": 0.069290214948921,
|
| 329 |
+
"learning_rate": 0.0004889547035820856,
|
| 330 |
+
"loss": 3.1689,
|
| 331 |
+
"step": 46
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.13,
|
| 335 |
+
"grad_norm": 0.06229058085155359,
|
| 336 |
+
"learning_rate": 0.0004883006150073028,
|
| 337 |
+
"loss": 3.1514,
|
| 338 |
+
"step": 47
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.13,
|
| 342 |
+
"grad_norm": 0.05042088665341411,
|
| 343 |
+
"learning_rate": 0.000487628175587465,
|
| 344 |
+
"loss": 3.1816,
|
| 345 |
+
"step": 48
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.13,
|
| 349 |
+
"grad_norm": 0.0545027205027528,
|
| 350 |
+
"learning_rate": 0.00048693743710519807,
|
| 351 |
+
"loss": 3.1875,
|
| 352 |
+
"step": 49
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.14,
|
| 356 |
+
"grad_norm": 0.06583086092960536,
|
| 357 |
+
"learning_rate": 0.00048622845275228614,
|
| 358 |
+
"loss": 3.1816,
|
| 359 |
+
"step": 50
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.14,
|
| 363 |
+
"grad_norm": 0.06171597888295892,
|
| 364 |
+
"learning_rate": 0.00048550127712557515,
|
| 365 |
+
"loss": 3.1621,
|
| 366 |
+
"step": 51
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.14,
|
| 370 |
+
"grad_norm": 0.05936904165187056,
|
| 371 |
+
"learning_rate": 0.0004847559662227691,
|
| 372 |
+
"loss": 3.1689,
|
| 373 |
+
"step": 52
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.14,
|
| 377 |
+
"grad_norm": 0.053617959255968184,
|
| 378 |
+
"learning_rate": 0.0004839925774381172,
|
| 379 |
+
"loss": 3.1826,
|
| 380 |
+
"step": 53
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.15,
|
| 384 |
+
"grad_norm": 0.0584940883576519,
|
| 385 |
+
"learning_rate": 0.00048321116955799426,
|
| 386 |
+
"loss": 3.1445,
|
| 387 |
+
"step": 54
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.15,
|
| 391 |
+
"grad_norm": 0.05534676948352036,
|
| 392 |
+
"learning_rate": 0.0004824118027563742,
|
| 393 |
+
"loss": 3.1367,
|
| 394 |
+
"step": 55
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.15,
|
| 398 |
+
"grad_norm": 0.05079053816193199,
|
| 399 |
+
"learning_rate": 0.00048159453859019535,
|
| 400 |
+
"loss": 3.1221,
|
| 401 |
+
"step": 56
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.15,
|
| 405 |
+
"grad_norm": 0.051247127284195455,
|
| 406 |
+
"learning_rate": 0.000480759439994621,
|
| 407 |
+
"loss": 3.1533,
|
| 408 |
+
"step": 57
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.16,
|
| 412 |
+
"grad_norm": 0.05840283966287256,
|
| 413 |
+
"learning_rate": 0.00047990657127819234,
|
| 414 |
+
"loss": 3.1592,
|
| 415 |
+
"step": 58
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.16,
|
| 419 |
+
"grad_norm": 0.06275081599011795,
|
| 420 |
+
"learning_rate": 0.0004790359981178766,
|
| 421 |
+
"loss": 3.1309,
|
| 422 |
+
"step": 59
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.16,
|
| 426 |
+
"grad_norm": 0.061480429010200704,
|
| 427 |
+
"learning_rate": 0.0004781477875540091,
|
| 428 |
+
"loss": 3.1484,
|
| 429 |
+
"step": 60
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.16,
|
| 433 |
+
"grad_norm": 0.06296237796551907,
|
| 434 |
+
"learning_rate": 0.0004772420079851313,
|
| 435 |
+
"loss": 3.1338,
|
| 436 |
+
"step": 61
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.17,
|
| 440 |
+
"grad_norm": 0.05939876669569374,
|
| 441 |
+
"learning_rate": 0.00047631872916272265,
|
| 442 |
+
"loss": 3.1758,
|
| 443 |
+
"step": 62
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.17,
|
| 447 |
+
"grad_norm": 0.05697500202970297,
|
| 448 |
+
"learning_rate": 0.00047537802218583036,
|
| 449 |
+
"loss": 3.1631,
|
| 450 |
+
"step": 63
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.17,
|
| 454 |
+
"grad_norm": 0.05235594093980006,
|
| 455 |
+
"learning_rate": 0.00047441995949559323,
|
| 456 |
+
"loss": 3.1592,
|
| 457 |
+
"step": 64
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.18,
|
| 461 |
+
"grad_norm": 0.05136699721805155,
|
| 462 |
+
"learning_rate": 0.0004734446148696641,
|
| 463 |
+
"loss": 3.1348,
|
| 464 |
+
"step": 65
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.18,
|
| 468 |
+
"grad_norm": 0.056572106168923254,
|
| 469 |
+
"learning_rate": 0.00047245206341652734,
|
| 470 |
+
"loss": 3.1475,
|
| 471 |
+
"step": 66
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.18,
|
| 475 |
+
"grad_norm": 0.0680276012617714,
|
| 476 |
+
"learning_rate": 0.0004714423815697162,
|
| 477 |
+
"loss": 3.1221,
|
| 478 |
+
"step": 67
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.18,
|
| 482 |
+
"grad_norm": 0.058203239195036796,
|
| 483 |
+
"learning_rate": 0.0004704156470819261,
|
| 484 |
+
"loss": 3.1387,
|
| 485 |
+
"step": 68
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.19,
|
| 489 |
+
"grad_norm": 0.05654758029632964,
|
| 490 |
+
"learning_rate": 0.00046937193901902703,
|
| 491 |
+
"loss": 3.1152,
|
| 492 |
+
"step": 69
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.19,
|
| 496 |
+
"grad_norm": 0.07079606102138672,
|
| 497 |
+
"learning_rate": 0.0004683113377539756,
|
| 498 |
+
"loss": 3.0811,
|
| 499 |
+
"step": 70
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.19,
|
| 503 |
+
"grad_norm": 0.06311116636541605,
|
| 504 |
+
"learning_rate": 0.00046723392496062513,
|
| 505 |
+
"loss": 3.1299,
|
| 506 |
+
"step": 71
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.19,
|
| 510 |
+
"grad_norm": 0.04930899192235485,
|
| 511 |
+
"learning_rate": 0.0004661397836074362,
|
| 512 |
+
"loss": 3.0898,
|
| 513 |
+
"step": 72
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.2,
|
| 517 |
+
"grad_norm": 0.07242568044263648,
|
| 518 |
+
"learning_rate": 0.0004650289979510883,
|
| 519 |
+
"loss": 3.1328,
|
| 520 |
+
"step": 73
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.2,
|
| 524 |
+
"grad_norm": 0.06815171525302943,
|
| 525 |
+
"learning_rate": 0.00046390165352998996,
|
| 526 |
+
"loss": 3.0713,
|
| 527 |
+
"step": 74
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.2,
|
| 531 |
+
"grad_norm": 0.06138422128904437,
|
| 532 |
+
"learning_rate": 0.0004627578371576933,
|
| 533 |
+
"loss": 3.0703,
|
| 534 |
+
"step": 75
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.21,
|
| 538 |
+
"grad_norm": 0.059644963475315685,
|
| 539 |
+
"learning_rate": 0.0004615976369162077,
|
| 540 |
+
"loss": 3.0898,
|
| 541 |
+
"step": 76
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.21,
|
| 545 |
+
"grad_norm": 0.06496043006168827,
|
| 546 |
+
"learning_rate": 0.00046042114214921723,
|
| 547 |
+
"loss": 3.0986,
|
| 548 |
+
"step": 77
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.21,
|
| 552 |
+
"grad_norm": 0.0634821846883854,
|
| 553 |
+
"learning_rate": 0.00045922844345520046,
|
| 554 |
+
"loss": 3.1025,
|
| 555 |
+
"step": 78
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.21,
|
| 559 |
+
"grad_norm": 0.06014574410746669,
|
| 560 |
+
"learning_rate": 0.000458019632680454,
|
| 561 |
+
"loss": 3.1055,
|
| 562 |
+
"step": 79
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.22,
|
| 566 |
+
"grad_norm": 0.06497852199883221,
|
| 567 |
+
"learning_rate": 0.00045679480291201916,
|
| 568 |
+
"loss": 3.0781,
|
| 569 |
+
"step": 80
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.22,
|
| 573 |
+
"grad_norm": 0.07113303653242763,
|
| 574 |
+
"learning_rate": 0.00045555404847051406,
|
| 575 |
+
"loss": 3.0742,
|
| 576 |
+
"step": 81
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.22,
|
| 580 |
+
"grad_norm": 0.06890895616862315,
|
| 581 |
+
"learning_rate": 0.0004542974649028701,
|
| 582 |
+
"loss": 3.0605,
|
| 583 |
+
"step": 82
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.22,
|
| 587 |
+
"grad_norm": 0.06609605331363849,
|
| 588 |
+
"learning_rate": 0.00045302514897497414,
|
| 589 |
+
"loss": 3.0859,
|
| 590 |
+
"step": 83
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.23,
|
| 594 |
+
"grad_norm": 0.06783983704880324,
|
| 595 |
+
"learning_rate": 0.0004517371986642168,
|
| 596 |
+
"loss": 3.0693,
|
| 597 |
+
"step": 84
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.23,
|
| 601 |
+
"grad_norm": 0.07647246769844311,
|
| 602 |
+
"learning_rate": 0.0004504337131519477,
|
| 603 |
+
"loss": 3.0918,
|
| 604 |
+
"step": 85
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.23,
|
| 608 |
+
"grad_norm": 0.07603324518310302,
|
| 609 |
+
"learning_rate": 0.00044911479281583747,
|
| 610 |
+
"loss": 3.0547,
|
| 611 |
+
"step": 86
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.23,
|
| 615 |
+
"grad_norm": 0.08035265561784409,
|
| 616 |
+
"learning_rate": 0.0004477805392221483,
|
| 617 |
+
"loss": 3.0547,
|
| 618 |
+
"step": 87
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.24,
|
| 622 |
+
"grad_norm": 0.07818094146866135,
|
| 623 |
+
"learning_rate": 0.00044643105511791236,
|
| 624 |
+
"loss": 3.0938,
|
| 625 |
+
"step": 88
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.24,
|
| 629 |
+
"grad_norm": 0.08705128277730868,
|
| 630 |
+
"learning_rate": 0.0004450664444230197,
|
| 631 |
+
"loss": 3.0811,
|
| 632 |
+
"step": 89
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.24,
|
| 636 |
+
"grad_norm": 0.08927687330837643,
|
| 637 |
+
"learning_rate": 0.00044368681222221553,
|
| 638 |
+
"loss": 3.0566,
|
| 639 |
+
"step": 90
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.25,
|
| 643 |
+
"grad_norm": 0.08533504666800651,
|
| 644 |
+
"learning_rate": 0.00044229226475700815,
|
| 645 |
+
"loss": 3.0889,
|
| 646 |
+
"step": 91
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.25,
|
| 650 |
+
"grad_norm": 0.08432813563363031,
|
| 651 |
+
"learning_rate": 0.0004408829094174872,
|
| 652 |
+
"loss": 3.0312,
|
| 653 |
+
"step": 92
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.25,
|
| 657 |
+
"grad_norm": 0.09460079385850353,
|
| 658 |
+
"learning_rate": 0.0004394588547340546,
|
| 659 |
+
"loss": 3.041,
|
| 660 |
+
"step": 93
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.25,
|
| 664 |
+
"grad_norm": 0.12021954624176756,
|
| 665 |
+
"learning_rate": 0.00043802021036906604,
|
| 666 |
+
"loss": 3.0293,
|
| 667 |
+
"step": 94
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.26,
|
| 671 |
+
"grad_norm": 0.09741352747719122,
|
| 672 |
+
"learning_rate": 0.00043656708710838687,
|
| 673 |
+
"loss": 3.0381,
|
| 674 |
+
"step": 95
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.26,
|
| 678 |
+
"grad_norm": 0.1005763508223488,
|
| 679 |
+
"learning_rate": 0.00043509959685286047,
|
| 680 |
+
"loss": 3.043,
|
| 681 |
+
"step": 96
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.26,
|
| 685 |
+
"grad_norm": 0.11242499386359363,
|
| 686 |
+
"learning_rate": 0.00043361785260969126,
|
| 687 |
+
"loss": 3.0674,
|
| 688 |
+
"step": 97
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.26,
|
| 692 |
+
"grad_norm": 0.10975775405702991,
|
| 693 |
+
"learning_rate": 0.0004321219684837422,
|
| 694 |
+
"loss": 3.0312,
|
| 695 |
+
"step": 98
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.27,
|
| 699 |
+
"grad_norm": 0.11737592618628766,
|
| 700 |
+
"learning_rate": 0.00043061205966874784,
|
| 701 |
+
"loss": 3.0547,
|
| 702 |
+
"step": 99
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.27,
|
| 706 |
+
"grad_norm": 0.13051148309222815,
|
| 707 |
+
"learning_rate": 0.00042908824243844383,
|
| 708 |
+
"loss": 3.0225,
|
| 709 |
+
"step": 100
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.27,
|
| 713 |
+
"grad_norm": 0.1293140100108899,
|
| 714 |
+
"learning_rate": 0.00042755063413761304,
|
| 715 |
+
"loss": 3.0352,
|
| 716 |
+
"step": 101
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.28,
|
| 720 |
+
"grad_norm": 0.11252955969439887,
|
| 721 |
+
"learning_rate": 0.0004259993531730486,
|
| 722 |
+
"loss": 3.0303,
|
| 723 |
+
"step": 102
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.28,
|
| 727 |
+
"grad_norm": 0.1328900419617491,
|
| 728 |
+
"learning_rate": 0.00042443451900443666,
|
| 729 |
+
"loss": 3.0146,
|
| 730 |
+
"step": 103
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.28,
|
| 734 |
+
"grad_norm": 0.12693126792922263,
|
| 735 |
+
"learning_rate": 0.0004228562521351562,
|
| 736 |
+
"loss": 3.0244,
|
| 737 |
+
"step": 104
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.28,
|
| 741 |
+
"grad_norm": 0.1251814736102281,
|
| 742 |
+
"learning_rate": 0.0004212646741030001,
|
| 743 |
+
"loss": 3.0264,
|
| 744 |
+
"step": 105
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.29,
|
| 748 |
+
"grad_norm": 0.13416576734949723,
|
| 749 |
+
"learning_rate": 0.0004196599074708156,
|
| 750 |
+
"loss": 3.0166,
|
| 751 |
+
"step": 106
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.29,
|
| 755 |
+
"grad_norm": 0.12552446417030455,
|
| 756 |
+
"learning_rate": 0.0004180420758170659,
|
| 757 |
+
"loss": 3.0332,
|
| 758 |
+
"step": 107
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.29,
|
| 762 |
+
"grad_norm": 0.13647763499691762,
|
| 763 |
+
"learning_rate": 0.00041641130372631417,
|
| 764 |
+
"loss": 3.0049,
|
| 765 |
+
"step": 108
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.29,
|
| 769 |
+
"grad_norm": 0.12497753797378944,
|
| 770 |
+
"learning_rate": 0.0004147677167796294,
|
| 771 |
+
"loss": 2.9766,
|
| 772 |
+
"step": 109
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.3,
|
| 776 |
+
"grad_norm": 0.16181335617568174,
|
| 777 |
+
"learning_rate": 0.00041311144154491567,
|
| 778 |
+
"loss": 3.0078,
|
| 779 |
+
"step": 110
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.3,
|
| 783 |
+
"grad_norm": 0.1231265891953706,
|
| 784 |
+
"learning_rate": 0.0004114426055671655,
|
| 785 |
+
"loss": 3.0371,
|
| 786 |
+
"step": 111
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.3,
|
| 790 |
+
"grad_norm": 0.1500040260456989,
|
| 791 |
+
"learning_rate": 0.0004097613373586385,
|
| 792 |
+
"loss": 3.0137,
|
| 793 |
+
"step": 112
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.31,
|
| 797 |
+
"grad_norm": 0.14493053467762143,
|
| 798 |
+
"learning_rate": 0.0004080677663889644,
|
| 799 |
+
"loss": 3.0059,
|
| 800 |
+
"step": 113
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.31,
|
| 804 |
+
"grad_norm": 0.13498422720011877,
|
| 805 |
+
"learning_rate": 0.00040636202307517313,
|
| 806 |
+
"loss": 2.9834,
|
| 807 |
+
"step": 114
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.31,
|
| 811 |
+
"grad_norm": 0.13758614368752517,
|
| 812 |
+
"learning_rate": 0.0004046442387716519,
|
| 813 |
+
"loss": 2.9883,
|
| 814 |
+
"step": 115
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.31,
|
| 818 |
+
"grad_norm": 0.1547910759873873,
|
| 819 |
+
"learning_rate": 0.0004029145457600298,
|
| 820 |
+
"loss": 3.0146,
|
| 821 |
+
"step": 116
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.32,
|
| 825 |
+
"grad_norm": 0.1114547142061138,
|
| 826 |
+
"learning_rate": 0.0004011730772389916,
|
| 827 |
+
"loss": 2.9502,
|
| 828 |
+
"step": 117
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.32,
|
| 832 |
+
"grad_norm": 0.158851085984572,
|
| 833 |
+
"learning_rate": 0.0003994199673140199,
|
| 834 |
+
"loss": 2.9707,
|
| 835 |
+
"step": 118
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.32,
|
| 839 |
+
"grad_norm": 0.15035124478352418,
|
| 840 |
+
"learning_rate": 0.0003976553509870683,
|
| 841 |
+
"loss": 2.959,
|
| 842 |
+
"step": 119
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.32,
|
| 846 |
+
"grad_norm": 0.18562089234260126,
|
| 847 |
+
"learning_rate": 0.00039587936414616556,
|
| 848 |
+
"loss": 2.9893,
|
| 849 |
+
"step": 120
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.33,
|
| 853 |
+
"grad_norm": 0.1500050006066447,
|
| 854 |
+
"learning_rate": 0.00039409214355495093,
|
| 855 |
+
"loss": 2.9912,
|
| 856 |
+
"step": 121
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.33,
|
| 860 |
+
"grad_norm": 0.15173547245397362,
|
| 861 |
+
"learning_rate": 0.00039229382684214233,
|
| 862 |
+
"loss": 2.9971,
|
| 863 |
+
"step": 122
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.33,
|
| 867 |
+
"grad_norm": 0.17021016573674075,
|
| 868 |
+
"learning_rate": 0.0003904845524909382,
|
| 869 |
+
"loss": 2.9766,
|
| 870 |
+
"step": 123
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.33,
|
| 874 |
+
"grad_norm": 0.16160325963256653,
|
| 875 |
+
"learning_rate": 0.0003886644598283533,
|
| 876 |
+
"loss": 2.9463,
|
| 877 |
+
"step": 124
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.34,
|
| 881 |
+
"grad_norm": 0.16620594275883835,
|
| 882 |
+
"learning_rate": 0.00038683368901448934,
|
| 883 |
+
"loss": 2.9951,
|
| 884 |
+
"step": 125
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.34,
|
| 888 |
+
"grad_norm": 0.18996567972484682,
|
| 889 |
+
"learning_rate": 0.00038499238103174166,
|
| 890 |
+
"loss": 2.9492,
|
| 891 |
+
"step": 126
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.34,
|
| 895 |
+
"grad_norm": 0.17307569767495753,
|
| 896 |
+
"learning_rate": 0.00038314067767394286,
|
| 897 |
+
"loss": 2.9805,
|
| 898 |
+
"step": 127
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.35,
|
| 902 |
+
"grad_norm": 0.19685088775869003,
|
| 903 |
+
"learning_rate": 0.0003812787215354433,
|
| 904 |
+
"loss": 2.9424,
|
| 905 |
+
"step": 128
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.35,
|
| 909 |
+
"grad_norm": 0.16189642626507808,
|
| 910 |
+
"learning_rate": 0.00037940665600013057,
|
| 911 |
+
"loss": 2.9512,
|
| 912 |
+
"step": 129
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.35,
|
| 916 |
+
"grad_norm": 0.17612947304733076,
|
| 917 |
+
"learning_rate": 0.00037752462523038793,
|
| 918 |
+
"loss": 2.9639,
|
| 919 |
+
"step": 130
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.35,
|
| 923 |
+
"grad_norm": 0.18179176164130736,
|
| 924 |
+
"learning_rate": 0.0003756327741559925,
|
| 925 |
+
"loss": 2.9688,
|
| 926 |
+
"step": 131
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.36,
|
| 930 |
+
"grad_norm": 0.16591267900078252,
|
| 931 |
+
"learning_rate": 0.00037373124846295505,
|
| 932 |
+
"loss": 2.9482,
|
| 933 |
+
"step": 132
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.36,
|
| 937 |
+
"grad_norm": 0.18798955074723858,
|
| 938 |
+
"learning_rate": 0.00037182019458230066,
|
| 939 |
+
"loss": 2.9395,
|
| 940 |
+
"step": 133
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.36,
|
| 944 |
+
"grad_norm": 0.17543591147515358,
|
| 945 |
+
"learning_rate": 0.000369899759678793,
|
| 946 |
+
"loss": 2.9404,
|
| 947 |
+
"step": 134
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.36,
|
| 951 |
+
"grad_norm": 0.18490619032970695,
|
| 952 |
+
"learning_rate": 0.0003679700916396009,
|
| 953 |
+
"loss": 2.9736,
|
| 954 |
+
"step": 135
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.37,
|
| 958 |
+
"grad_norm": 0.1804894673337707,
|
| 959 |
+
"learning_rate": 0.00036603133906291095,
|
| 960 |
+
"loss": 2.9561,
|
| 961 |
+
"step": 136
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.37,
|
| 965 |
+
"grad_norm": 0.18154491542277257,
|
| 966 |
+
"learning_rate": 0.0003640836512464837,
|
| 967 |
+
"loss": 2.9609,
|
| 968 |
+
"step": 137
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.37,
|
| 972 |
+
"grad_norm": 0.15353929430307406,
|
| 973 |
+
"learning_rate": 0.00036212717817615635,
|
| 974 |
+
"loss": 2.9355,
|
| 975 |
+
"step": 138
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.38,
|
| 979 |
+
"grad_norm": 0.21621902507983504,
|
| 980 |
+
"learning_rate": 0.000360162070514294,
|
| 981 |
+
"loss": 2.957,
|
| 982 |
+
"step": 139
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.38,
|
| 986 |
+
"grad_norm": 0.1829941353882955,
|
| 987 |
+
"learning_rate": 0.0003581884795881861,
|
| 988 |
+
"loss": 2.9521,
|
| 989 |
+
"step": 140
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.38,
|
| 993 |
+
"grad_norm": 0.18716180371474261,
|
| 994 |
+
"learning_rate": 0.0003562065573783945,
|
| 995 |
+
"loss": 2.9434,
|
| 996 |
+
"step": 141
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.38,
|
| 1000 |
+
"grad_norm": 0.2275322416310555,
|
| 1001 |
+
"learning_rate": 0.0003542164565070486,
|
| 1002 |
+
"loss": 2.9365,
|
| 1003 |
+
"step": 142
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.39,
|
| 1007 |
+
"grad_norm": 0.19125954171713724,
|
| 1008 |
+
"learning_rate": 0.00035221833022609366,
|
| 1009 |
+
"loss": 2.9121,
|
| 1010 |
+
"step": 143
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.39,
|
| 1014 |
+
"grad_norm": 0.23978782914992927,
|
| 1015 |
+
"learning_rate": 0.0003502123324054882,
|
| 1016 |
+
"loss": 2.9473,
|
| 1017 |
+
"step": 144
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.39,
|
| 1021 |
+
"grad_norm": 0.18956138126965327,
|
| 1022 |
+
"learning_rate": 0.0003481986175213556,
|
| 1023 |
+
"loss": 2.9531,
|
| 1024 |
+
"step": 145
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.39,
|
| 1028 |
+
"grad_norm": 0.20333907762683046,
|
| 1029 |
+
"learning_rate": 0.0003461773406440879,
|
| 1030 |
+
"loss": 2.9229,
|
| 1031 |
+
"step": 146
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.4,
|
| 1035 |
+
"grad_norm": 0.21650596272683872,
|
| 1036 |
+
"learning_rate": 0.0003441486574264048,
|
| 1037 |
+
"loss": 2.9639,
|
| 1038 |
+
"step": 147
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.4,
|
| 1042 |
+
"grad_norm": 0.20068147784646773,
|
| 1043 |
+
"learning_rate": 0.00034211272409136667,
|
| 1044 |
+
"loss": 2.9395,
|
| 1045 |
+
"step": 148
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.4,
|
| 1049 |
+
"grad_norm": 0.20053669008991853,
|
| 1050 |
+
"learning_rate": 0.00034006969742034465,
|
| 1051 |
+
"loss": 2.9541,
|
| 1052 |
+
"step": 149
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.41,
|
| 1056 |
+
"grad_norm": 0.19123221320191458,
|
| 1057 |
+
"learning_rate": 0.00033801973474094745,
|
| 1058 |
+
"loss": 2.9668,
|
| 1059 |
+
"step": 150
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.41,
|
| 1063 |
+
"grad_norm": 0.20055890299268714,
|
| 1064 |
+
"learning_rate": 0.00033596299391490565,
|
| 1065 |
+
"loss": 2.918,
|
| 1066 |
+
"step": 151
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.41,
|
| 1070 |
+
"grad_norm": 0.21961462852023858,
|
| 1071 |
+
"learning_rate": 0.0003338996333259155,
|
| 1072 |
+
"loss": 2.8887,
|
| 1073 |
+
"step": 152
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.41,
|
| 1077 |
+
"grad_norm": 0.19923388725160776,
|
| 1078 |
+
"learning_rate": 0.000331829811867442,
|
| 1079 |
+
"loss": 2.915,
|
| 1080 |
+
"step": 153
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.42,
|
| 1084 |
+
"grad_norm": 0.2202467020445118,
|
| 1085 |
+
"learning_rate": 0.0003297536889304835,
|
| 1086 |
+
"loss": 2.9072,
|
| 1087 |
+
"step": 154
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.42,
|
| 1091 |
+
"grad_norm": 0.1896310412062489,
|
| 1092 |
+
"learning_rate": 0.0003276714243912966,
|
| 1093 |
+
"loss": 2.9219,
|
| 1094 |
+
"step": 155
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.42,
|
| 1098 |
+
"grad_norm": 0.24063275517469532,
|
| 1099 |
+
"learning_rate": 0.0003255831785990854,
|
| 1100 |
+
"loss": 2.9268,
|
| 1101 |
+
"step": 156
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.42,
|
| 1105 |
+
"grad_norm": 0.2093352162026971,
|
| 1106 |
+
"learning_rate": 0.000323489112363653,
|
| 1107 |
+
"loss": 2.9102,
|
| 1108 |
+
"step": 157
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.43,
|
| 1112 |
+
"grad_norm": 0.19084865959587785,
|
| 1113 |
+
"learning_rate": 0.000321389386943018,
|
| 1114 |
+
"loss": 2.916,
|
| 1115 |
+
"step": 158
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.43,
|
| 1119 |
+
"grad_norm": 0.24890451630400445,
|
| 1120 |
+
"learning_rate": 0.0003192841640309966,
|
| 1121 |
+
"loss": 2.9492,
|
| 1122 |
+
"step": 159
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.43,
|
| 1126 |
+
"grad_norm": 0.18002995253459828,
|
| 1127 |
+
"learning_rate": 0.0003171736057447511,
|
| 1128 |
+
"loss": 2.9111,
|
| 1129 |
+
"step": 160
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.43,
|
| 1133 |
+
"grad_norm": 0.22447733600257028,
|
| 1134 |
+
"learning_rate": 0.00031505787461230553,
|
| 1135 |
+
"loss": 2.8936,
|
| 1136 |
+
"step": 161
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.44,
|
| 1140 |
+
"grad_norm": 0.2467901168622779,
|
| 1141 |
+
"learning_rate": 0.00031293713356003,
|
| 1142 |
+
"loss": 2.9316,
|
| 1143 |
+
"step": 162
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.44,
|
| 1147 |
+
"grad_norm": 0.20550840515489216,
|
| 1148 |
+
"learning_rate": 0.00031081154590009404,
|
| 1149 |
+
"loss": 2.9453,
|
| 1150 |
+
"step": 163
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.44,
|
| 1154 |
+
"grad_norm": 0.21166602849404573,
|
| 1155 |
+
"learning_rate": 0.0003086812753178907,
|
| 1156 |
+
"loss": 2.916,
|
| 1157 |
+
"step": 164
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.45,
|
| 1161 |
+
"grad_norm": 0.3602799328889876,
|
| 1162 |
+
"learning_rate": 0.0003065464858594311,
|
| 1163 |
+
"loss": 2.9102,
|
| 1164 |
+
"step": 165
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.45,
|
| 1168 |
+
"grad_norm": 0.18298535380931488,
|
| 1169 |
+
"learning_rate": 0.00030440734191871234,
|
| 1170 |
+
"loss": 2.9043,
|
| 1171 |
+
"step": 166
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.45,
|
| 1175 |
+
"grad_norm": 0.2572700716773119,
|
| 1176 |
+
"learning_rate": 0.00030226400822505736,
|
| 1177 |
+
"loss": 2.9033,
|
| 1178 |
+
"step": 167
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.45,
|
| 1182 |
+
"grad_norm": 0.25835419996657133,
|
| 1183 |
+
"learning_rate": 0.00030011664983043025,
|
| 1184 |
+
"loss": 2.9062,
|
| 1185 |
+
"step": 168
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.46,
|
| 1189 |
+
"grad_norm": 0.17683277759979313,
|
| 1190 |
+
"learning_rate": 0.0002979654320967255,
|
| 1191 |
+
"loss": 2.916,
|
| 1192 |
+
"step": 169
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.46,
|
| 1196 |
+
"grad_norm": 0.2894378726624145,
|
| 1197 |
+
"learning_rate": 0.0002958105206830343,
|
| 1198 |
+
"loss": 2.9053,
|
| 1199 |
+
"step": 170
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.46,
|
| 1203 |
+
"grad_norm": 0.23685763825987682,
|
| 1204 |
+
"learning_rate": 0.0002936520815328876,
|
| 1205 |
+
"loss": 2.8857,
|
| 1206 |
+
"step": 171
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.46,
|
| 1210 |
+
"grad_norm": 0.2008573100452373,
|
| 1211 |
+
"learning_rate": 0.000291490280861477,
|
| 1212 |
+
"loss": 2.9297,
|
| 1213 |
+
"step": 172
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.47,
|
| 1217 |
+
"grad_norm": 0.26444755166927053,
|
| 1218 |
+
"learning_rate": 0.00028932528514285534,
|
| 1219 |
+
"loss": 2.9316,
|
| 1220 |
+
"step": 173
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.47,
|
| 1224 |
+
"grad_norm": 0.24379542789804157,
|
| 1225 |
+
"learning_rate": 0.00028715726109711665,
|
| 1226 |
+
"loss": 2.9131,
|
| 1227 |
+
"step": 174
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.47,
|
| 1231 |
+
"grad_norm": 0.20673216784596224,
|
| 1232 |
+
"learning_rate": 0.00028498637567755817,
|
| 1233 |
+
"loss": 2.9082,
|
| 1234 |
+
"step": 175
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.48,
|
| 1238 |
+
"grad_norm": 0.21944969615374402,
|
| 1239 |
+
"learning_rate": 0.0002828127960578227,
|
| 1240 |
+
"loss": 2.9199,
|
| 1241 |
+
"step": 176
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.48,
|
| 1245 |
+
"grad_norm": 0.20478834301866145,
|
| 1246 |
+
"learning_rate": 0.00028063668961902623,
|
| 1247 |
+
"loss": 2.9033,
|
| 1248 |
+
"step": 177
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 0.48,
|
| 1252 |
+
"grad_norm": 0.21481876173786985,
|
| 1253 |
+
"learning_rate": 0.00027845822393686724,
|
| 1254 |
+
"loss": 2.8809,
|
| 1255 |
+
"step": 178
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.48,
|
| 1259 |
+
"grad_norm": 0.22811931240082148,
|
| 1260 |
+
"learning_rate": 0.00027627756676872344,
|
| 1261 |
+
"loss": 2.8916,
|
| 1262 |
+
"step": 179
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.49,
|
| 1266 |
+
"grad_norm": 0.25707311204201183,
|
| 1267 |
+
"learning_rate": 0.00027409488604073184,
|
| 1268 |
+
"loss": 2.9258,
|
| 1269 |
+
"step": 180
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.49,
|
| 1273 |
+
"grad_norm": 0.22655991900141323,
|
| 1274 |
+
"learning_rate": 0.0002719103498348586,
|
| 1275 |
+
"loss": 2.8936,
|
| 1276 |
+
"step": 181
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.49,
|
| 1280 |
+
"grad_norm": 0.2462832484307767,
|
| 1281 |
+
"learning_rate": 0.00026972412637595445,
|
| 1282 |
+
"loss": 2.8789,
|
| 1283 |
+
"step": 182
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.49,
|
| 1287 |
+
"grad_norm": 0.16995673551677068,
|
| 1288 |
+
"learning_rate": 0.00026753638401880094,
|
| 1289 |
+
"loss": 2.876,
|
| 1290 |
+
"step": 183
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.5,
|
| 1294 |
+
"grad_norm": 0.2536950503045075,
|
| 1295 |
+
"learning_rate": 0.0002653472912351454,
|
| 1296 |
+
"loss": 2.9189,
|
| 1297 |
+
"step": 184
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 0.5,
|
| 1301 |
+
"grad_norm": 0.20069386810955497,
|
| 1302 |
+
"learning_rate": 0.0002631570166007276,
|
| 1303 |
+
"loss": 2.8838,
|
| 1304 |
+
"step": 185
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.5,
|
| 1308 |
+
"grad_norm": 0.17449546614470834,
|
| 1309 |
+
"learning_rate": 0.00026096572878229855,
|
| 1310 |
+
"loss": 2.8936,
|
| 1311 |
+
"step": 186
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.51,
|
| 1315 |
+
"grad_norm": 0.20654114988441505,
|
| 1316 |
+
"learning_rate": 0.00025877359652463124,
|
| 1317 |
+
"loss": 2.8965,
|
| 1318 |
+
"step": 187
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 0.51,
|
| 1322 |
+
"grad_norm": 0.19593843356526874,
|
| 1323 |
+
"learning_rate": 0.0002565807886375267,
|
| 1324 |
+
"loss": 2.8984,
|
| 1325 |
+
"step": 188
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.51,
|
| 1329 |
+
"grad_norm": 0.20292351001271516,
|
| 1330 |
+
"learning_rate": 0.00025438747398281433,
|
| 1331 |
+
"loss": 2.9004,
|
| 1332 |
+
"step": 189
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 0.51,
|
| 1336 |
+
"grad_norm": 0.2043498399445655,
|
| 1337 |
+
"learning_rate": 0.00025219382146134806,
|
| 1338 |
+
"loss": 2.8516,
|
| 1339 |
+
"step": 190
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 0.52,
|
| 1343 |
+
"grad_norm": 0.1661636250320572,
|
| 1344 |
+
"learning_rate": 0.00025,
|
| 1345 |
+
"loss": 2.8545,
|
| 1346 |
+
"step": 191
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.52,
|
| 1350 |
+
"grad_norm": 0.19022198317016406,
|
| 1351 |
+
"learning_rate": 0.00024780617853865195,
|
| 1352 |
+
"loss": 2.9092,
|
| 1353 |
+
"step": 192
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.52,
|
| 1357 |
+
"grad_norm": 0.22844383852438707,
|
| 1358 |
+
"learning_rate": 0.0002456125260171857,
|
| 1359 |
+
"loss": 2.9111,
|
| 1360 |
+
"step": 193
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.52,
|
| 1364 |
+
"grad_norm": 0.17676802172340947,
|
| 1365 |
+
"learning_rate": 0.00024341921136247338,
|
| 1366 |
+
"loss": 2.8857,
|
| 1367 |
+
"step": 194
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.53,
|
| 1371 |
+
"grad_norm": 0.17406205780039047,
|
| 1372 |
+
"learning_rate": 0.00024122640347536885,
|
| 1373 |
+
"loss": 2.8896,
|
| 1374 |
+
"step": 195
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.53,
|
| 1378 |
+
"grad_norm": 0.21596180880748264,
|
| 1379 |
+
"learning_rate": 0.00023903427121770148,
|
| 1380 |
+
"loss": 2.8643,
|
| 1381 |
+
"step": 196
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 0.53,
|
| 1385 |
+
"grad_norm": 0.2302727016974318,
|
| 1386 |
+
"learning_rate": 0.0002368429833992724,
|
| 1387 |
+
"loss": 2.8936,
|
| 1388 |
+
"step": 197
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.53,
|
| 1392 |
+
"grad_norm": 0.193768899062196,
|
| 1393 |
+
"learning_rate": 0.00023465270876485474,
|
| 1394 |
+
"loss": 2.8496,
|
| 1395 |
+
"step": 198
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.54,
|
| 1399 |
+
"grad_norm": 0.1784013570789268,
|
| 1400 |
+
"learning_rate": 0.0002324636159811991,
|
| 1401 |
+
"loss": 2.8379,
|
| 1402 |
+
"step": 199
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.54,
|
| 1406 |
+
"grad_norm": 0.20783430292995747,
|
| 1407 |
+
"learning_rate": 0.00023027587362404558,
|
| 1408 |
+
"loss": 2.8799,
|
| 1409 |
+
"step": 200
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 0.54,
|
| 1413 |
+
"grad_norm": 0.20944704580582277,
|
| 1414 |
+
"learning_rate": 0.00022808965016514143,
|
| 1415 |
+
"loss": 2.8965,
|
| 1416 |
+
"step": 201
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 0.55,
|
| 1420 |
+
"grad_norm": 0.21496187154467147,
|
| 1421 |
+
"learning_rate": 0.00022590511395926811,
|
| 1422 |
+
"loss": 2.8672,
|
| 1423 |
+
"step": 202
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 0.55,
|
| 1427 |
+
"grad_norm": 0.21633737213344018,
|
| 1428 |
+
"learning_rate": 0.00022372243323127663,
|
| 1429 |
+
"loss": 2.875,
|
| 1430 |
+
"step": 203
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.55,
|
| 1434 |
+
"grad_norm": 0.19865551825102143,
|
| 1435 |
+
"learning_rate": 0.00022154177606313277,
|
| 1436 |
+
"loss": 2.8672,
|
| 1437 |
+
"step": 204
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.55,
|
| 1441 |
+
"grad_norm": 0.18963996378916614,
|
| 1442 |
+
"learning_rate": 0.00021936331038097386,
|
| 1443 |
+
"loss": 2.9023,
|
| 1444 |
+
"step": 205
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 0.56,
|
| 1448 |
+
"grad_norm": 0.20425004554589565,
|
| 1449 |
+
"learning_rate": 0.00021718720394217725,
|
| 1450 |
+
"loss": 2.8574,
|
| 1451 |
+
"step": 206
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 0.56,
|
| 1455 |
+
"grad_norm": 0.2555973427913448,
|
| 1456 |
+
"learning_rate": 0.00021501362432244193,
|
| 1457 |
+
"loss": 2.8701,
|
| 1458 |
+
"step": 207
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 0.56,
|
| 1462 |
+
"grad_norm": 0.1953962081104273,
|
| 1463 |
+
"learning_rate": 0.00021284273890288336,
|
| 1464 |
+
"loss": 2.8779,
|
| 1465 |
+
"step": 208
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.56,
|
| 1469 |
+
"grad_norm": 0.2248900981224502,
|
| 1470 |
+
"learning_rate": 0.00021067471485714472,
|
| 1471 |
+
"loss": 2.8398,
|
| 1472 |
+
"step": 209
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.57,
|
| 1476 |
+
"grad_norm": 0.26993244613823164,
|
| 1477 |
+
"learning_rate": 0.00020850971913852307,
|
| 1478 |
+
"loss": 2.8613,
|
| 1479 |
+
"step": 210
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.57,
|
| 1483 |
+
"grad_norm": 0.2022265920119534,
|
| 1484 |
+
"learning_rate": 0.00020634791846711243,
|
| 1485 |
+
"loss": 2.8574,
|
| 1486 |
+
"step": 211
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 0.57,
|
| 1490 |
+
"grad_norm": 0.24664106252564721,
|
| 1491 |
+
"learning_rate": 0.00020418947931696575,
|
| 1492 |
+
"loss": 2.8564,
|
| 1493 |
+
"step": 212
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 0.58,
|
| 1497 |
+
"grad_norm": 0.1967992968536126,
|
| 1498 |
+
"learning_rate": 0.00020203456790327453,
|
| 1499 |
+
"loss": 2.8555,
|
| 1500 |
+
"step": 213
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 0.58,
|
| 1504 |
+
"grad_norm": 0.22710937800094108,
|
| 1505 |
+
"learning_rate": 0.0001998833501695698,
|
| 1506 |
+
"loss": 2.8594,
|
| 1507 |
+
"step": 214
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.58,
|
| 1511 |
+
"grad_norm": 0.2609592105615878,
|
| 1512 |
+
"learning_rate": 0.00019773599177494265,
|
| 1513 |
+
"loss": 2.8682,
|
| 1514 |
+
"step": 215
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.58,
|
| 1518 |
+
"grad_norm": 0.20377916758575193,
|
| 1519 |
+
"learning_rate": 0.00019559265808128778,
|
| 1520 |
+
"loss": 2.8496,
|
| 1521 |
+
"step": 216
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.59,
|
| 1525 |
+
"grad_norm": 0.2298131342082367,
|
| 1526 |
+
"learning_rate": 0.00019345351414056894,
|
| 1527 |
+
"loss": 2.8398,
|
| 1528 |
+
"step": 217
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 0.59,
|
| 1532 |
+
"grad_norm": 0.234771024880693,
|
| 1533 |
+
"learning_rate": 0.00019131872468210938,
|
| 1534 |
+
"loss": 2.8467,
|
| 1535 |
+
"step": 218
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 0.59,
|
| 1539 |
+
"grad_norm": 0.23844110318408976,
|
| 1540 |
+
"learning_rate": 0.00018918845409990592,
|
| 1541 |
+
"loss": 2.917,
|
| 1542 |
+
"step": 219
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 0.59,
|
| 1546 |
+
"grad_norm": 0.19036660800238525,
|
| 1547 |
+
"learning_rate": 0.00018706286643996997,
|
| 1548 |
+
"loss": 2.8564,
|
| 1549 |
+
"step": 220
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 0.6,
|
| 1553 |
+
"grad_norm": 0.20622165918252797,
|
| 1554 |
+
"learning_rate": 0.0001849421253876945,
|
| 1555 |
+
"loss": 2.8848,
|
| 1556 |
+
"step": 221
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.6,
|
| 1560 |
+
"grad_norm": 0.1883236349273076,
|
| 1561 |
+
"learning_rate": 0.000182826394255249,
|
| 1562 |
+
"loss": 2.8643,
|
| 1563 |
+
"step": 222
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.6,
|
| 1567 |
+
"grad_norm": 0.22937231306412564,
|
| 1568 |
+
"learning_rate": 0.00018071583596900343,
|
| 1569 |
+
"loss": 2.8623,
|
| 1570 |
+
"step": 223
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 0.6,
|
| 1574 |
+
"grad_norm": 0.21538020857939882,
|
| 1575 |
+
"learning_rate": 0.00017861061305698205,
|
| 1576 |
+
"loss": 2.8438,
|
| 1577 |
+
"step": 224
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 0.61,
|
| 1581 |
+
"grad_norm": 0.1918870218278195,
|
| 1582 |
+
"learning_rate": 0.0001765108876363471,
|
| 1583 |
+
"loss": 2.8467,
|
| 1584 |
+
"step": 225
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 0.61,
|
| 1588 |
+
"grad_norm": 0.21091693539511583,
|
| 1589 |
+
"learning_rate": 0.00017441682140091464,
|
| 1590 |
+
"loss": 2.8281,
|
| 1591 |
+
"step": 226
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 0.61,
|
| 1595 |
+
"grad_norm": 0.30476893084177187,
|
| 1596 |
+
"learning_rate": 0.0001723285756087034,
|
| 1597 |
+
"loss": 2.873,
|
| 1598 |
+
"step": 227
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.62,
|
| 1602 |
+
"grad_norm": 0.17816221080916358,
|
| 1603 |
+
"learning_rate": 0.0001702463110695166,
|
| 1604 |
+
"loss": 2.8525,
|
| 1605 |
+
"step": 228
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.62,
|
| 1609 |
+
"grad_norm": 0.195493340516346,
|
| 1610 |
+
"learning_rate": 0.00016817018813255796,
|
| 1611 |
+
"loss": 2.8359,
|
| 1612 |
+
"step": 229
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 0.62,
|
| 1616 |
+
"grad_norm": 0.23565534271641125,
|
| 1617 |
+
"learning_rate": 0.00016610036667408462,
|
| 1618 |
+
"loss": 2.8535,
|
| 1619 |
+
"step": 230
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 0.62,
|
| 1623 |
+
"grad_norm": 0.2033674101509759,
|
| 1624 |
+
"learning_rate": 0.00016403700608509438,
|
| 1625 |
+
"loss": 2.832,
|
| 1626 |
+
"step": 231
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 0.63,
|
| 1630 |
+
"grad_norm": 0.1643046553266504,
|
| 1631 |
+
"learning_rate": 0.00016198026525905256,
|
| 1632 |
+
"loss": 2.8525,
|
| 1633 |
+
"step": 232
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 0.63,
|
| 1637 |
+
"grad_norm": 0.18680871871908325,
|
| 1638 |
+
"learning_rate": 0.00015993030257965536,
|
| 1639 |
+
"loss": 2.8496,
|
| 1640 |
+
"step": 233
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.63,
|
| 1644 |
+
"grad_norm": 0.21879472472240535,
|
| 1645 |
+
"learning_rate": 0.00015788727590863345,
|
| 1646 |
+
"loss": 2.8457,
|
| 1647 |
+
"step": 234
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.63,
|
| 1651 |
+
"grad_norm": 0.189459681090247,
|
| 1652 |
+
"learning_rate": 0.00015585134257359523,
|
| 1653 |
+
"loss": 2.8252,
|
| 1654 |
+
"step": 235
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 0.64,
|
| 1658 |
+
"grad_norm": 0.20698851721297937,
|
| 1659 |
+
"learning_rate": 0.00015382265935591212,
|
| 1660 |
+
"loss": 2.8545,
|
| 1661 |
+
"step": 236
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 0.64,
|
| 1665 |
+
"grad_norm": 0.21230475441950047,
|
| 1666 |
+
"learning_rate": 0.00015180138247864445,
|
| 1667 |
+
"loss": 2.8682,
|
| 1668 |
+
"step": 237
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 0.64,
|
| 1672 |
+
"grad_norm": 0.2170131840363288,
|
| 1673 |
+
"learning_rate": 0.00014978766759451187,
|
| 1674 |
+
"loss": 2.835,
|
| 1675 |
+
"step": 238
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 0.65,
|
| 1679 |
+
"grad_norm": 0.1747044380388941,
|
| 1680 |
+
"learning_rate": 0.00014778166977390643,
|
| 1681 |
+
"loss": 2.8193,
|
| 1682 |
+
"step": 239
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 0.65,
|
| 1686 |
+
"grad_norm": 0.2195091552174841,
|
| 1687 |
+
"learning_rate": 0.00014578354349295138,
|
| 1688 |
+
"loss": 2.8428,
|
| 1689 |
+
"step": 240
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 0.65,
|
| 1693 |
+
"grad_norm": 0.17839712623841394,
|
| 1694 |
+
"learning_rate": 0.0001437934426216056,
|
| 1695 |
+
"loss": 2.8496,
|
| 1696 |
+
"step": 241
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 0.65,
|
| 1700 |
+
"grad_norm": 0.2218268901944312,
|
| 1701 |
+
"learning_rate": 0.0001418115204118139,
|
| 1702 |
+
"loss": 2.8379,
|
| 1703 |
+
"step": 242
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 0.66,
|
| 1707 |
+
"grad_norm": 0.22291238199967986,
|
| 1708 |
+
"learning_rate": 0.00013983792948570606,
|
| 1709 |
+
"loss": 2.8418,
|
| 1710 |
+
"step": 243
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 0.66,
|
| 1714 |
+
"grad_norm": 0.2314705382253311,
|
| 1715 |
+
"learning_rate": 0.00013787282182384363,
|
| 1716 |
+
"loss": 2.835,
|
| 1717 |
+
"step": 244
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 0.66,
|
| 1721 |
+
"grad_norm": 0.2062566807258234,
|
| 1722 |
+
"learning_rate": 0.0001359163487535164,
|
| 1723 |
+
"loss": 2.8174,
|
| 1724 |
+
"step": 245
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 0.66,
|
| 1728 |
+
"grad_norm": 0.20906998538613936,
|
| 1729 |
+
"learning_rate": 0.00013396866093708898,
|
| 1730 |
+
"loss": 2.8184,
|
| 1731 |
+
"step": 246
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 0.67,
|
| 1735 |
+
"grad_norm": 0.2687448764467888,
|
| 1736 |
+
"learning_rate": 0.0001320299083603992,
|
| 1737 |
+
"loss": 2.835,
|
| 1738 |
+
"step": 247
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.67,
|
| 1742 |
+
"grad_norm": 0.1990426441612401,
|
| 1743 |
+
"learning_rate": 0.00013010024032120718,
|
| 1744 |
+
"loss": 2.877,
|
| 1745 |
+
"step": 248
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 0.67,
|
| 1749 |
+
"grad_norm": 0.20269787106945644,
|
| 1750 |
+
"learning_rate": 0.00012817980541769935,
|
| 1751 |
+
"loss": 2.8564,
|
| 1752 |
+
"step": 249
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 0.68,
|
| 1756 |
+
"grad_norm": 0.19805581453093807,
|
| 1757 |
+
"learning_rate": 0.000126268751537045,
|
| 1758 |
+
"loss": 2.8447,
|
| 1759 |
+
"step": 250
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 0.68,
|
| 1763 |
+
"grad_norm": 0.204837890537741,
|
| 1764 |
+
"learning_rate": 0.00012436722584400751,
|
| 1765 |
+
"loss": 2.8418,
|
| 1766 |
+
"step": 251
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 0.68,
|
| 1770 |
+
"grad_norm": 0.169050938877304,
|
| 1771 |
+
"learning_rate": 0.00012247537476961214,
|
| 1772 |
+
"loss": 2.8311,
|
| 1773 |
+
"step": 252
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 0.68,
|
| 1777 |
+
"grad_norm": 0.21928573521752834,
|
| 1778 |
+
"learning_rate": 0.00012059334399986949,
|
| 1779 |
+
"loss": 2.8066,
|
| 1780 |
+
"step": 253
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 0.69,
|
| 1784 |
+
"grad_norm": 0.18055284433976476,
|
| 1785 |
+
"learning_rate": 0.00011872127846455672,
|
| 1786 |
+
"loss": 2.8125,
|
| 1787 |
+
"step": 254
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 0.69,
|
| 1791 |
+
"grad_norm": 0.20659092646221383,
|
| 1792 |
+
"learning_rate": 0.00011685932232605718,
|
| 1793 |
+
"loss": 2.8066,
|
| 1794 |
+
"step": 255
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 0.69,
|
| 1798 |
+
"grad_norm": 0.20299552954457484,
|
| 1799 |
+
"learning_rate": 0.00011500761896825842,
|
| 1800 |
+
"loss": 2.8398,
|
| 1801 |
+
"step": 256
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 0.69,
|
| 1805 |
+
"grad_norm": 0.21714725029231918,
|
| 1806 |
+
"learning_rate": 0.00011316631098551067,
|
| 1807 |
+
"loss": 2.832,
|
| 1808 |
+
"step": 257
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 0.7,
|
| 1812 |
+
"grad_norm": 0.21963153956688958,
|
| 1813 |
+
"learning_rate": 0.00011133554017164671,
|
| 1814 |
+
"loss": 2.8193,
|
| 1815 |
+
"step": 258
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 0.7,
|
| 1819 |
+
"grad_norm": 0.19726411127227286,
|
| 1820 |
+
"learning_rate": 0.00010951544750906181,
|
| 1821 |
+
"loss": 2.8213,
|
| 1822 |
+
"step": 259
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 0.7,
|
| 1826 |
+
"grad_norm": 0.21949146356088167,
|
| 1827 |
+
"learning_rate": 0.00010770617315785766,
|
| 1828 |
+
"loss": 2.8418,
|
| 1829 |
+
"step": 260
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 0.7,
|
| 1833 |
+
"grad_norm": 0.2126404293875657,
|
| 1834 |
+
"learning_rate": 0.00010590785644504919,
|
| 1835 |
+
"loss": 2.833,
|
| 1836 |
+
"step": 261
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 0.71,
|
| 1840 |
+
"grad_norm": 0.2071663035323768,
|
| 1841 |
+
"learning_rate": 0.00010412063585383452,
|
| 1842 |
+
"loss": 2.8174,
|
| 1843 |
+
"step": 262
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 0.71,
|
| 1847 |
+
"grad_norm": 0.222005661427221,
|
| 1848 |
+
"learning_rate": 0.00010234464901293173,
|
| 1849 |
+
"loss": 2.8076,
|
| 1850 |
+
"step": 263
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.71,
|
| 1854 |
+
"grad_norm": 0.2233059412032165,
|
| 1855 |
+
"learning_rate": 0.00010058003268598015,
|
| 1856 |
+
"loss": 2.8389,
|
| 1857 |
+
"step": 264
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 0.72,
|
| 1861 |
+
"grad_norm": 0.20234119110348878,
|
| 1862 |
+
"learning_rate": 9.882692276100841e-05,
|
| 1863 |
+
"loss": 2.8242,
|
| 1864 |
+
"step": 265
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 0.72,
|
| 1868 |
+
"grad_norm": 0.22629912766899482,
|
| 1869 |
+
"learning_rate": 9.70854542399702e-05,
|
| 1870 |
+
"loss": 2.8232,
|
| 1871 |
+
"step": 266
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 0.72,
|
| 1875 |
+
"grad_norm": 0.19123353391124664,
|
| 1876 |
+
"learning_rate": 9.535576122834819e-05,
|
| 1877 |
+
"loss": 2.8418,
|
| 1878 |
+
"step": 267
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 0.72,
|
| 1882 |
+
"grad_norm": 0.20149824977959827,
|
| 1883 |
+
"learning_rate": 9.363797692482695e-05,
|
| 1884 |
+
"loss": 2.8115,
|
| 1885 |
+
"step": 268
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 0.73,
|
| 1889 |
+
"grad_norm": 0.17961635360231348,
|
| 1890 |
+
"learning_rate": 9.193223361103556e-05,
|
| 1891 |
+
"loss": 2.8291,
|
| 1892 |
+
"step": 269
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 0.73,
|
| 1896 |
+
"grad_norm": 0.2121856691428259,
|
| 1897 |
+
"learning_rate": 9.023866264136154e-05,
|
| 1898 |
+
"loss": 2.8457,
|
| 1899 |
+
"step": 270
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 0.73,
|
| 1903 |
+
"grad_norm": 0.18851239383885068,
|
| 1904 |
+
"learning_rate": 8.85573944328345e-05,
|
| 1905 |
+
"loss": 2.7979,
|
| 1906 |
+
"step": 271
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 0.73,
|
| 1910 |
+
"grad_norm": 0.17521924172534067,
|
| 1911 |
+
"learning_rate": 8.688855845508442e-05,
|
| 1912 |
+
"loss": 2.8086,
|
| 1913 |
+
"step": 272
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 0.74,
|
| 1917 |
+
"grad_norm": 0.18640071329178245,
|
| 1918 |
+
"learning_rate": 8.523228322037063e-05,
|
| 1919 |
+
"loss": 2.8369,
|
| 1920 |
+
"step": 273
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 0.74,
|
| 1924 |
+
"grad_norm": 0.16959390971032334,
|
| 1925 |
+
"learning_rate": 8.358869627368584e-05,
|
| 1926 |
+
"loss": 2.8398,
|
| 1927 |
+
"step": 274
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 0.74,
|
| 1931 |
+
"grad_norm": 0.21111788917507904,
|
| 1932 |
+
"learning_rate": 8.195792418293416e-05,
|
| 1933 |
+
"loss": 2.7852,
|
| 1934 |
+
"step": 275
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 0.75,
|
| 1938 |
+
"grad_norm": 0.17179812719144402,
|
| 1939 |
+
"learning_rate": 8.03400925291845e-05,
|
| 1940 |
+
"loss": 2.8115,
|
| 1941 |
+
"step": 276
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 0.75,
|
| 1945 |
+
"grad_norm": 0.20491937385204168,
|
| 1946 |
+
"learning_rate": 7.873532589699988e-05,
|
| 1947 |
+
"loss": 2.791,
|
| 1948 |
+
"step": 277
|
| 1949 |
+
},
|
| 1950 |
+
{
|
| 1951 |
+
"epoch": 0.75,
|
| 1952 |
+
"grad_norm": 0.16357238878820704,
|
| 1953 |
+
"learning_rate": 7.71437478648438e-05,
|
| 1954 |
+
"loss": 2.8291,
|
| 1955 |
+
"step": 278
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 0.75,
|
| 1959 |
+
"grad_norm": 0.1835990750271033,
|
| 1960 |
+
"learning_rate": 7.556548099556338e-05,
|
| 1961 |
+
"loss": 2.7998,
|
| 1962 |
+
"step": 279
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 0.76,
|
| 1966 |
+
"grad_norm": 0.1907507274896681,
|
| 1967 |
+
"learning_rate": 7.40006468269514e-05,
|
| 1968 |
+
"loss": 2.8213,
|
| 1969 |
+
"step": 280
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 0.76,
|
| 1973 |
+
"grad_norm": 0.18564884527181547,
|
| 1974 |
+
"learning_rate": 7.244936586238704e-05,
|
| 1975 |
+
"loss": 2.8301,
|
| 1976 |
+
"step": 281
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 0.76,
|
| 1980 |
+
"grad_norm": 0.17700032167669388,
|
| 1981 |
+
"learning_rate": 7.091175756155618e-05,
|
| 1982 |
+
"loss": 2.8115,
|
| 1983 |
+
"step": 282
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 0.76,
|
| 1987 |
+
"grad_norm": 0.16847559163220463,
|
| 1988 |
+
"learning_rate": 6.938794033125225e-05,
|
| 1989 |
+
"loss": 2.833,
|
| 1990 |
+
"step": 283
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 0.77,
|
| 1994 |
+
"grad_norm": 0.18723560047102772,
|
| 1995 |
+
"learning_rate": 6.787803151625791e-05,
|
| 1996 |
+
"loss": 2.8281,
|
| 1997 |
+
"step": 284
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 0.77,
|
| 2001 |
+
"grad_norm": 0.18633675001733044,
|
| 2002 |
+
"learning_rate": 6.63821473903087e-05,
|
| 2003 |
+
"loss": 2.8184,
|
| 2004 |
+
"step": 285
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 0.77,
|
| 2008 |
+
"grad_norm": 0.1822056204472567,
|
| 2009 |
+
"learning_rate": 6.490040314713952e-05,
|
| 2010 |
+
"loss": 2.834,
|
| 2011 |
+
"step": 286
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 0.78,
|
| 2015 |
+
"grad_norm": 0.19871618333174051,
|
| 2016 |
+
"learning_rate": 6.343291289161316e-05,
|
| 2017 |
+
"loss": 2.8174,
|
| 2018 |
+
"step": 287
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 0.78,
|
| 2022 |
+
"grad_norm": 0.20546173536543255,
|
| 2023 |
+
"learning_rate": 6.197978963093401e-05,
|
| 2024 |
+
"loss": 2.8242,
|
| 2025 |
+
"step": 288
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 0.78,
|
| 2029 |
+
"grad_norm": 0.19374612820702158,
|
| 2030 |
+
"learning_rate": 6.054114526594548e-05,
|
| 2031 |
+
"loss": 2.835,
|
| 2032 |
+
"step": 289
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 0.78,
|
| 2036 |
+
"grad_norm": 0.18924589658433175,
|
| 2037 |
+
"learning_rate": 5.911709058251283e-05,
|
| 2038 |
+
"loss": 2.7988,
|
| 2039 |
+
"step": 290
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 0.79,
|
| 2043 |
+
"grad_norm": 0.22644302618589385,
|
| 2044 |
+
"learning_rate": 5.770773524299192e-05,
|
| 2045 |
+
"loss": 2.8525,
|
| 2046 |
+
"step": 291
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 0.79,
|
| 2050 |
+
"grad_norm": 0.18156812551515011,
|
| 2051 |
+
"learning_rate": 5.631318777778452e-05,
|
| 2052 |
+
"loss": 2.8047,
|
| 2053 |
+
"step": 292
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 0.79,
|
| 2057 |
+
"grad_norm": 0.19435635715822475,
|
| 2058 |
+
"learning_rate": 5.493355557698029e-05,
|
| 2059 |
+
"loss": 2.8203,
|
| 2060 |
+
"step": 293
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 0.79,
|
| 2064 |
+
"grad_norm": 0.20414237163125332,
|
| 2065 |
+
"learning_rate": 5.3568944882087604e-05,
|
| 2066 |
+
"loss": 2.8369,
|
| 2067 |
+
"step": 294
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 0.8,
|
| 2071 |
+
"grad_norm": 0.18567184675458911,
|
| 2072 |
+
"learning_rate": 5.2219460777851695e-05,
|
| 2073 |
+
"loss": 2.8398,
|
| 2074 |
+
"step": 295
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 0.8,
|
| 2078 |
+
"grad_norm": 0.1783969629345075,
|
| 2079 |
+
"learning_rate": 5.088520718416254e-05,
|
| 2080 |
+
"loss": 2.8057,
|
| 2081 |
+
"step": 296
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 0.8,
|
| 2085 |
+
"grad_norm": 0.18247845423738968,
|
| 2086 |
+
"learning_rate": 4.9566286848052363e-05,
|
| 2087 |
+
"loss": 2.835,
|
| 2088 |
+
"step": 297
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 0.8,
|
| 2092 |
+
"grad_norm": 0.18785911029833752,
|
| 2093 |
+
"learning_rate": 4.826280133578323e-05,
|
| 2094 |
+
"loss": 2.8242,
|
| 2095 |
+
"step": 298
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 0.81,
|
| 2099 |
+
"grad_norm": 0.18953403976774194,
|
| 2100 |
+
"learning_rate": 4.697485102502588e-05,
|
| 2101 |
+
"loss": 2.8252,
|
| 2102 |
+
"step": 299
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 0.81,
|
| 2106 |
+
"grad_norm": 0.18190067624900982,
|
| 2107 |
+
"learning_rate": 4.5702535097129906e-05,
|
| 2108 |
+
"loss": 2.835,
|
| 2109 |
+
"step": 300
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 0.81,
|
| 2113 |
+
"grad_norm": 0.20251912717942616,
|
| 2114 |
+
"learning_rate": 4.444595152948605e-05,
|
| 2115 |
+
"loss": 2.8242,
|
| 2116 |
+
"step": 301
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 0.82,
|
| 2120 |
+
"grad_norm": 0.18960132345131622,
|
| 2121 |
+
"learning_rate": 4.32051970879809e-05,
|
| 2122 |
+
"loss": 2.8418,
|
| 2123 |
+
"step": 302
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 0.82,
|
| 2127 |
+
"grad_norm": 0.20472286168933038,
|
| 2128 |
+
"learning_rate": 4.198036731954605e-05,
|
| 2129 |
+
"loss": 2.79,
|
| 2130 |
+
"step": 303
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 0.82,
|
| 2134 |
+
"grad_norm": 0.1885255817179192,
|
| 2135 |
+
"learning_rate": 4.077155654479955e-05,
|
| 2136 |
+
"loss": 2.8076,
|
| 2137 |
+
"step": 304
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 0.82,
|
| 2141 |
+
"grad_norm": 0.17197377072124265,
|
| 2142 |
+
"learning_rate": 3.957885785078277e-05,
|
| 2143 |
+
"loss": 2.8125,
|
| 2144 |
+
"step": 305
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 0.83,
|
| 2148 |
+
"grad_norm": 0.17243813839090957,
|
| 2149 |
+
"learning_rate": 3.840236308379233e-05,
|
| 2150 |
+
"loss": 2.7998,
|
| 2151 |
+
"step": 306
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 0.83,
|
| 2155 |
+
"grad_norm": 0.187479341506351,
|
| 2156 |
+
"learning_rate": 3.7242162842306724e-05,
|
| 2157 |
+
"loss": 2.8076,
|
| 2158 |
+
"step": 307
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 0.83,
|
| 2162 |
+
"grad_norm": 0.19619960697754343,
|
| 2163 |
+
"learning_rate": 3.609834647001001e-05,
|
| 2164 |
+
"loss": 2.8125,
|
| 2165 |
+
"step": 308
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 0.83,
|
| 2169 |
+
"grad_norm": 0.1969308587109652,
|
| 2170 |
+
"learning_rate": 3.497100204891177e-05,
|
| 2171 |
+
"loss": 2.7891,
|
| 2172 |
+
"step": 309
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 0.84,
|
| 2176 |
+
"grad_norm": 0.1856744385771311,
|
| 2177 |
+
"learning_rate": 3.386021639256373e-05,
|
| 2178 |
+
"loss": 2.8076,
|
| 2179 |
+
"step": 310
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 0.84,
|
| 2183 |
+
"grad_norm": 0.19472326340578114,
|
| 2184 |
+
"learning_rate": 3.2766075039374926e-05,
|
| 2185 |
+
"loss": 2.8135,
|
| 2186 |
+
"step": 311
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 0.84,
|
| 2190 |
+
"grad_norm": 0.16977810395348839,
|
| 2191 |
+
"learning_rate": 3.168866224602443e-05,
|
| 2192 |
+
"loss": 2.8008,
|
| 2193 |
+
"step": 312
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 0.85,
|
| 2197 |
+
"grad_norm": 0.16737273719653672,
|
| 2198 |
+
"learning_rate": 3.0628060980972956e-05,
|
| 2199 |
+
"loss": 2.8096,
|
| 2200 |
+
"step": 313
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 0.85,
|
| 2204 |
+
"grad_norm": 0.18706543053983507,
|
| 2205 |
+
"learning_rate": 2.958435291807396e-05,
|
| 2206 |
+
"loss": 2.8223,
|
| 2207 |
+
"step": 314
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 0.85,
|
| 2211 |
+
"grad_norm": 0.20138299122195422,
|
| 2212 |
+
"learning_rate": 2.8557618430283845e-05,
|
| 2213 |
+
"loss": 2.8145,
|
| 2214 |
+
"step": 315
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 0.85,
|
| 2218 |
+
"grad_norm": 0.1825577738379651,
|
| 2219 |
+
"learning_rate": 2.7547936583472682e-05,
|
| 2220 |
+
"loss": 2.8135,
|
| 2221 |
+
"step": 316
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 0.86,
|
| 2225 |
+
"grad_norm": 0.1681442717695743,
|
| 2226 |
+
"learning_rate": 2.655538513033598e-05,
|
| 2227 |
+
"loss": 2.8174,
|
| 2228 |
+
"step": 317
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 0.86,
|
| 2232 |
+
"grad_norm": 0.1886334015299787,
|
| 2233 |
+
"learning_rate": 2.5580040504406736e-05,
|
| 2234 |
+
"loss": 2.8105,
|
| 2235 |
+
"step": 318
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 0.86,
|
| 2239 |
+
"grad_norm": 0.18472284304096284,
|
| 2240 |
+
"learning_rate": 2.4621977814169668e-05,
|
| 2241 |
+
"loss": 2.8076,
|
| 2242 |
+
"step": 319
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 0.86,
|
| 2246 |
+
"grad_norm": 0.16272769219517882,
|
| 2247 |
+
"learning_rate": 2.3681270837277365e-05,
|
| 2248 |
+
"loss": 2.8174,
|
| 2249 |
+
"step": 320
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 0.87,
|
| 2253 |
+
"grad_norm": 0.19968898028239582,
|
| 2254 |
+
"learning_rate": 2.2757992014868806e-05,
|
| 2255 |
+
"loss": 2.8242,
|
| 2256 |
+
"step": 321
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 0.87,
|
| 2260 |
+
"grad_norm": 0.17043048449106335,
|
| 2261 |
+
"learning_rate": 2.1852212445990894e-05,
|
| 2262 |
+
"loss": 2.8135,
|
| 2263 |
+
"step": 322
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 0.87,
|
| 2267 |
+
"grad_norm": 0.1644894369897433,
|
| 2268 |
+
"learning_rate": 2.0964001882123466e-05,
|
| 2269 |
+
"loss": 2.7998,
|
| 2270 |
+
"step": 323
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 0.88,
|
| 2274 |
+
"grad_norm": 0.16881143272811544,
|
| 2275 |
+
"learning_rate": 2.0093428721807694e-05,
|
| 2276 |
+
"loss": 2.8252,
|
| 2277 |
+
"step": 324
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 0.88,
|
| 2281 |
+
"grad_norm": 0.16555621080410832,
|
| 2282 |
+
"learning_rate": 1.924056000537905e-05,
|
| 2283 |
+
"loss": 2.7939,
|
| 2284 |
+
"step": 325
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 0.88,
|
| 2288 |
+
"grad_norm": 0.16784278084821697,
|
| 2289 |
+
"learning_rate": 1.8405461409804676e-05,
|
| 2290 |
+
"loss": 2.8242,
|
| 2291 |
+
"step": 326
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 0.88,
|
| 2295 |
+
"grad_norm": 0.1641280362316012,
|
| 2296 |
+
"learning_rate": 1.7588197243625816e-05,
|
| 2297 |
+
"loss": 2.8223,
|
| 2298 |
+
"step": 327
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 0.89,
|
| 2302 |
+
"grad_norm": 0.17345327239016983,
|
| 2303 |
+
"learning_rate": 1.6788830442005742e-05,
|
| 2304 |
+
"loss": 2.835,
|
| 2305 |
+
"step": 328
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 0.89,
|
| 2309 |
+
"grad_norm": 0.1678603553402573,
|
| 2310 |
+
"learning_rate": 1.600742256188287e-05,
|
| 2311 |
+
"loss": 2.8623,
|
| 2312 |
+
"step": 329
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 0.89,
|
| 2316 |
+
"grad_norm": 0.17995413273432045,
|
| 2317 |
+
"learning_rate": 1.5244033777230909e-05,
|
| 2318 |
+
"loss": 2.7979,
|
| 2319 |
+
"step": 330
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 0.89,
|
| 2323 |
+
"grad_norm": 0.17259350839299145,
|
| 2324 |
+
"learning_rate": 1.4498722874424859e-05,
|
| 2325 |
+
"loss": 2.8105,
|
| 2326 |
+
"step": 331
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 0.9,
|
| 2330 |
+
"grad_norm": 0.1741787680290243,
|
| 2331 |
+
"learning_rate": 1.3771547247713956e-05,
|
| 2332 |
+
"loss": 2.7979,
|
| 2333 |
+
"step": 332
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 0.9,
|
| 2337 |
+
"grad_norm": 0.1654260393599542,
|
| 2338 |
+
"learning_rate": 1.3062562894801955e-05,
|
| 2339 |
+
"loss": 2.7998,
|
| 2340 |
+
"step": 333
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 0.9,
|
| 2344 |
+
"grad_norm": 0.15975325006455904,
|
| 2345 |
+
"learning_rate": 1.237182441253501e-05,
|
| 2346 |
+
"loss": 2.8076,
|
| 2347 |
+
"step": 334
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 0.9,
|
| 2351 |
+
"grad_norm": 0.1831835291092947,
|
| 2352 |
+
"learning_rate": 1.169938499269721e-05,
|
| 2353 |
+
"loss": 2.833,
|
| 2354 |
+
"step": 335
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 0.91,
|
| 2358 |
+
"grad_norm": 0.17054992749801087,
|
| 2359 |
+
"learning_rate": 1.1045296417914408e-05,
|
| 2360 |
+
"loss": 2.8125,
|
| 2361 |
+
"step": 336
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 0.91,
|
| 2365 |
+
"grad_norm": 0.1914560652459072,
|
| 2366 |
+
"learning_rate": 1.040960905766683e-05,
|
| 2367 |
+
"loss": 2.8174,
|
| 2368 |
+
"step": 337
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 0.91,
|
| 2372 |
+
"grad_norm": 0.16125789541336907,
|
| 2373 |
+
"learning_rate": 9.792371864409844e-06,
|
| 2374 |
+
"loss": 2.8145,
|
| 2375 |
+
"step": 338
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 0.92,
|
| 2379 |
+
"grad_norm": 0.18343996045358885,
|
| 2380 |
+
"learning_rate": 9.19363236980475e-06,
|
| 2381 |
+
"loss": 2.8105,
|
| 2382 |
+
"step": 339
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 0.92,
|
| 2386 |
+
"grad_norm": 0.19474642229296382,
|
| 2387 |
+
"learning_rate": 8.613436681058128e-06,
|
| 2388 |
+
"loss": 2.7979,
|
| 2389 |
+
"step": 340
|
| 2390 |
+
},
|
| 2391 |
+
{
|
| 2392 |
+
"epoch": 0.92,
|
| 2393 |
+
"grad_norm": 0.1541708717690255,
|
| 2394 |
+
"learning_rate": 8.051829477371458e-06,
|
| 2395 |
+
"loss": 2.8076,
|
| 2396 |
+
"step": 341
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 0.92,
|
| 2400 |
+
"grad_norm": 0.15274058388659703,
|
| 2401 |
+
"learning_rate": 7.508854006500393e-06,
|
| 2402 |
+
"loss": 2.8057,
|
| 2403 |
+
"step": 342
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 0.93,
|
| 2407 |
+
"grad_norm": 0.1778349122094768,
|
| 2408 |
+
"learning_rate": 6.984552081424456e-06,
|
| 2409 |
+
"loss": 2.8252,
|
| 2410 |
+
"step": 343
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 0.93,
|
| 2414 |
+
"grad_norm": 0.18535400361130824,
|
| 2415 |
+
"learning_rate": 6.4789640771270864e-06,
|
| 2416 |
+
"loss": 2.8506,
|
| 2417 |
+
"step": 344
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 0.93,
|
| 2421 |
+
"grad_norm": 0.17517598706340834,
|
| 2422 |
+
"learning_rate": 5.992128927486573e-06,
|
| 2423 |
+
"loss": 2.8047,
|
| 2424 |
+
"step": 345
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 0.93,
|
| 2428 |
+
"grad_norm": 0.1594767650558485,
|
| 2429 |
+
"learning_rate": 5.524084122277756e-06,
|
| 2430 |
+
"loss": 2.7871,
|
| 2431 |
+
"step": 346
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 0.94,
|
| 2435 |
+
"grad_norm": 0.17674478758195836,
|
| 2436 |
+
"learning_rate": 5.074865704284975e-06,
|
| 2437 |
+
"loss": 2.8242,
|
| 2438 |
+
"step": 347
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 0.94,
|
| 2442 |
+
"grad_norm": 0.1795205762704063,
|
| 2443 |
+
"learning_rate": 4.644508266526848e-06,
|
| 2444 |
+
"loss": 2.8037,
|
| 2445 |
+
"step": 348
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 0.94,
|
| 2449 |
+
"grad_norm": 0.16428283866890742,
|
| 2450 |
+
"learning_rate": 4.23304494959198e-06,
|
| 2451 |
+
"loss": 2.7852,
|
| 2452 |
+
"step": 349
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 0.95,
|
| 2456 |
+
"grad_norm": 0.1637667775135852,
|
| 2457 |
+
"learning_rate": 3.840507439087093e-06,
|
| 2458 |
+
"loss": 2.8252,
|
| 2459 |
+
"step": 350
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 0.95,
|
| 2463 |
+
"grad_norm": 0.1904128210839155,
|
| 2464 |
+
"learning_rate": 3.4669259631970873e-06,
|
| 2465 |
+
"loss": 2.8047,
|
| 2466 |
+
"step": 351
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 0.95,
|
| 2470 |
+
"grad_norm": 0.17893501780149162,
|
| 2471 |
+
"learning_rate": 3.1123292903569854e-06,
|
| 2472 |
+
"loss": 2.8115,
|
| 2473 |
+
"step": 352
|
| 2474 |
+
},
|
| 2475 |
+
{
|
| 2476 |
+
"epoch": 0.95,
|
| 2477 |
+
"grad_norm": 0.17860019081470324,
|
| 2478 |
+
"learning_rate": 2.7767447270368453e-06,
|
| 2479 |
+
"loss": 2.8184,
|
| 2480 |
+
"step": 353
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 0.96,
|
| 2484 |
+
"grad_norm": 0.16537005007226285,
|
| 2485 |
+
"learning_rate": 2.4601981156388566e-06,
|
| 2486 |
+
"loss": 2.7988,
|
| 2487 |
+
"step": 354
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 0.96,
|
| 2491 |
+
"grad_norm": 0.1482593636447568,
|
| 2492 |
+
"learning_rate": 2.162713832507157e-06,
|
| 2493 |
+
"loss": 2.7979,
|
| 2494 |
+
"step": 355
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 0.96,
|
| 2498 |
+
"grad_norm": 0.17394817869790966,
|
| 2499 |
+
"learning_rate": 1.8843147860509445e-06,
|
| 2500 |
+
"loss": 2.7988,
|
| 2501 |
+
"step": 356
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 0.96,
|
| 2505 |
+
"grad_norm": 0.1728142685313597,
|
| 2506 |
+
"learning_rate": 1.625022414980165e-06,
|
| 2507 |
+
"loss": 2.792,
|
| 2508 |
+
"step": 357
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 0.97,
|
| 2512 |
+
"grad_norm": 0.17603370515257444,
|
| 2513 |
+
"learning_rate": 1.3848566866545842e-06,
|
| 2514 |
+
"loss": 2.7998,
|
| 2515 |
+
"step": 358
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 0.97,
|
| 2519 |
+
"grad_norm": 0.16591885986335803,
|
| 2520 |
+
"learning_rate": 1.1638360955463222e-06,
|
| 2521 |
+
"loss": 2.8135,
|
| 2522 |
+
"step": 359
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 0.97,
|
| 2526 |
+
"grad_norm": 0.17699991948505095,
|
| 2527 |
+
"learning_rate": 9.619776618154386e-07,
|
| 2528 |
+
"loss": 2.8076,
|
| 2529 |
+
"step": 360
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 0.98,
|
| 2533 |
+
"grad_norm": 0.1690937483022535,
|
| 2534 |
+
"learning_rate": 7.792969299994512e-07,
|
| 2535 |
+
"loss": 2.8438,
|
| 2536 |
+
"step": 361
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 0.98,
|
| 2540 |
+
"grad_norm": 0.18098457876096985,
|
| 2541 |
+
"learning_rate": 6.158079678161566e-07,
|
| 2542 |
+
"loss": 2.8408,
|
| 2543 |
+
"step": 362
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 0.98,
|
| 2547 |
+
"grad_norm": 0.18213680457305967,
|
| 2548 |
+
"learning_rate": 4.7152336508038407e-07,
|
| 2549 |
+
"loss": 2.8311,
|
| 2550 |
+
"step": 363
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 0.98,
|
| 2554 |
+
"grad_norm": 0.16488850320672738,
|
| 2555 |
+
"learning_rate": 3.464542327344111e-07,
|
| 2556 |
+
"loss": 2.7949,
|
| 2557 |
+
"step": 364
|
| 2558 |
+
},
|
| 2559 |
+
{
|
| 2560 |
+
"epoch": 0.99,
|
| 2561 |
+
"grad_norm": 0.1606825195560946,
|
| 2562 |
+
"learning_rate": 2.406102019924805e-07,
|
| 2563 |
+
"loss": 2.8154,
|
| 2564 |
+
"step": 365
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 0.99,
|
| 2568 |
+
"grad_norm": 0.1743249393559749,
|
| 2569 |
+
"learning_rate": 1.5399942359897723e-07,
|
| 2570 |
+
"loss": 2.8174,
|
| 2571 |
+
"step": 366
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 0.99,
|
| 2575 |
+
"grad_norm": 0.1890660779935612,
|
| 2576 |
+
"learning_rate": 8.662856720090261e-08,
|
| 2577 |
+
"loss": 2.8359,
|
| 2578 |
+
"step": 367
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 0.99,
|
| 2582 |
+
"grad_norm": 0.17906338951545656,
|
| 2583 |
+
"learning_rate": 3.8502820834174226e-08,
|
| 2584 |
+
"loss": 2.7842,
|
| 2585 |
+
"step": 368
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 1.0,
|
| 2589 |
+
"grad_norm": 0.189945263964402,
|
| 2590 |
+
"learning_rate": 9.625890524084246e-09,
|
| 2591 |
+
"loss": 2.8037,
|
| 2592 |
+
"step": 369
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 1.0,
|
| 2596 |
+
"grad_norm": 0.16386427942822598,
|
| 2597 |
+
"learning_rate": 0.0,
|
| 2598 |
+
"loss": 2.8037,
|
| 2599 |
+
"step": 370
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 1.0,
|
| 2603 |
+
"step": 370,
|
| 2604 |
+
"total_flos": 2.715544613577294e+18,
|
| 2605 |
+
"train_loss": 3.01240234375,
|
| 2606 |
+
"train_runtime": 27753.3023,
|
| 2607 |
+
"train_samples_per_second": 27.316,
|
| 2608 |
+
"train_steps_per_second": 0.013
|
| 2609 |
+
}
|
| 2610 |
+
],
|
| 2611 |
+
"logging_steps": 1.0,
|
| 2612 |
+
"max_steps": 370,
|
| 2613 |
+
"num_input_tokens_seen": 0,
|
| 2614 |
+
"num_train_epochs": 1,
|
| 2615 |
+
"save_steps": 50,
|
| 2616 |
+
"total_flos": 2.715544613577294e+18,
|
| 2617 |
+
"train_batch_size": 128,
|
| 2618 |
+
"trial_name": null,
|
| 2619 |
+
"trial_params": null
|
| 2620 |
+
}
|