Upload 8 files
Browse files- config.json +160 -0
- generation_config.json +7 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +643 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,160 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"Idefics3ForConditionalGeneration"
|
| 4 |
+
],
|
| 5 |
+
"dtype": "float16",
|
| 6 |
+
"image_seq_len": 81,
|
| 7 |
+
"image_token_id": 49153,
|
| 8 |
+
"model_type": "idefics3",
|
| 9 |
+
"pad_token_id": 128002,
|
| 10 |
+
"scale_factor": 3,
|
| 11 |
+
"text_config": {
|
| 12 |
+
"_attn_implementation_autoset": false,
|
| 13 |
+
"_flash_attn_2_enabled": true,
|
| 14 |
+
"_name_or_path": "/fsx/m4/experiments/local_experiment_dir/s3_async_temporary_checkpoint_folder/tr_324_opt_400/unwrapped_model",
|
| 15 |
+
"architectures": [
|
| 16 |
+
"VLlama3ForCausalLM"
|
| 17 |
+
],
|
| 18 |
+
"attention_bias": false,
|
| 19 |
+
"attention_dropout": 0.0,
|
| 20 |
+
"bos_token_id": 0,
|
| 21 |
+
"dtype": "float16",
|
| 22 |
+
"eos_token_id": 0,
|
| 23 |
+
"head_dim": 64,
|
| 24 |
+
"hidden_act": "silu",
|
| 25 |
+
"hidden_size": 2048,
|
| 26 |
+
"initializer_range": 0.02,
|
| 27 |
+
"intermediate_size": 8192,
|
| 28 |
+
"max_position_embeddings": 16384,
|
| 29 |
+
"mlp_bias": false,
|
| 30 |
+
"model_type": "llama",
|
| 31 |
+
"neftune_noise_alpha": 0.0,
|
| 32 |
+
"num_attention_heads": 32,
|
| 33 |
+
"num_hidden_layers": 24,
|
| 34 |
+
"num_key_value_heads": 32,
|
| 35 |
+
"pad_token_id": 2,
|
| 36 |
+
"perceiver_config": {
|
| 37 |
+
"_attn_implementation_autoset": false,
|
| 38 |
+
"_name_or_path": "",
|
| 39 |
+
"add_cross_attention": false,
|
| 40 |
+
"architectures": null,
|
| 41 |
+
"attention_dropout": 0.0,
|
| 42 |
+
"bad_words_ids": null,
|
| 43 |
+
"begin_suppress_tokens": null,
|
| 44 |
+
"bos_token_id": null,
|
| 45 |
+
"chunk_size_feed_forward": 0,
|
| 46 |
+
"cross_attention_hidden_size": null,
|
| 47 |
+
"decoder_start_token_id": null,
|
| 48 |
+
"diversity_penalty": 0.0,
|
| 49 |
+
"do_sample": false,
|
| 50 |
+
"early_stopping": false,
|
| 51 |
+
"encoder_no_repeat_ngram_size": 0,
|
| 52 |
+
"eos_token_id": null,
|
| 53 |
+
"exponential_decay_length_penalty": null,
|
| 54 |
+
"finetuning_task": null,
|
| 55 |
+
"forced_bos_token_id": null,
|
| 56 |
+
"forced_eos_token_id": null,
|
| 57 |
+
"hidden_act": "silu",
|
| 58 |
+
"id2label": {
|
| 59 |
+
"0": "LABEL_0",
|
| 60 |
+
"1": "LABEL_1"
|
| 61 |
+
},
|
| 62 |
+
"is_decoder": false,
|
| 63 |
+
"is_encoder_decoder": false,
|
| 64 |
+
"label2id": {
|
| 65 |
+
"LABEL_0": 0,
|
| 66 |
+
"LABEL_1": 1
|
| 67 |
+
},
|
| 68 |
+
"length_penalty": 1.0,
|
| 69 |
+
"max_length": 20,
|
| 70 |
+
"min_length": 0,
|
| 71 |
+
"model_type": "vllama3",
|
| 72 |
+
"no_repeat_ngram_size": 0,
|
| 73 |
+
"num_beam_groups": 1,
|
| 74 |
+
"num_beams": 1,
|
| 75 |
+
"num_key_value_heads": 1,
|
| 76 |
+
"num_return_sequences": 1,
|
| 77 |
+
"output_attentions": false,
|
| 78 |
+
"output_hidden_states": false,
|
| 79 |
+
"output_scores": false,
|
| 80 |
+
"pad_token_id": null,
|
| 81 |
+
"prefix": null,
|
| 82 |
+
"problem_type": null,
|
| 83 |
+
"pruned_heads": {},
|
| 84 |
+
"qk_layer_norms_perceiver": false,
|
| 85 |
+
"remove_invalid_values": false,
|
| 86 |
+
"repetition_penalty": 1.0,
|
| 87 |
+
"resampler_depth": 6,
|
| 88 |
+
"resampler_head_dim": 96,
|
| 89 |
+
"resampler_n_heads": 16,
|
| 90 |
+
"resampler_n_latents": 64,
|
| 91 |
+
"return_dict": true,
|
| 92 |
+
"return_dict_in_generate": false,
|
| 93 |
+
"sep_token_id": null,
|
| 94 |
+
"suppress_tokens": null,
|
| 95 |
+
"task_specific_params": null,
|
| 96 |
+
"temperature": 1.0,
|
| 97 |
+
"tf_legacy_loss": false,
|
| 98 |
+
"tie_encoder_decoder": false,
|
| 99 |
+
"tie_word_embeddings": true,
|
| 100 |
+
"tokenizer_class": null,
|
| 101 |
+
"top_k": 50,
|
| 102 |
+
"top_p": 1.0,
|
| 103 |
+
"torch_dtype": null,
|
| 104 |
+
"torchscript": false,
|
| 105 |
+
"transformers_version": "4.46.0",
|
| 106 |
+
"typical_p": 1.0,
|
| 107 |
+
"use_bfloat16": false
|
| 108 |
+
},
|
| 109 |
+
"pretraining_tp": 1,
|
| 110 |
+
"qk_layer_norms": false,
|
| 111 |
+
"rms_norm_eps": 1e-05,
|
| 112 |
+
"rope_scaling": null,
|
| 113 |
+
"rope_theta": 273768.0,
|
| 114 |
+
"use_cache": true,
|
| 115 |
+
"use_resampler": false,
|
| 116 |
+
"vocab_size": 49155
|
| 117 |
+
},
|
| 118 |
+
"tie_word_embeddings": false,
|
| 119 |
+
"transformers.js_config": {
|
| 120 |
+
"dtype": {
|
| 121 |
+
"decoder_model_merged": "q4",
|
| 122 |
+
"embed_tokens": "auto",
|
| 123 |
+
"vision_encoder": "auto"
|
| 124 |
+
},
|
| 125 |
+
"kv_cache_dtype": {
|
| 126 |
+
"fp16": "float16",
|
| 127 |
+
"q4f16": "float16"
|
| 128 |
+
},
|
| 129 |
+
"use_external_data_format": {
|
| 130 |
+
"decoder_model_merged.onnx": true,
|
| 131 |
+
"decoder_model_merged_fp16.onnx": true
|
| 132 |
+
}
|
| 133 |
+
},
|
| 134 |
+
"transformers_version": "4.57.0",
|
| 135 |
+
"use_cache": true,
|
| 136 |
+
"vision_config": {
|
| 137 |
+
"_attn_implementation_autoset": false,
|
| 138 |
+
"attention_dropout": 0.0,
|
| 139 |
+
"dtype": "float16",
|
| 140 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 141 |
+
"hidden_size": 1152,
|
| 142 |
+
"image_size": 384,
|
| 143 |
+
"initializer_range": 0.02,
|
| 144 |
+
"intermediate_size": 4304,
|
| 145 |
+
"layer_norm_eps": 1e-06,
|
| 146 |
+
"max_image_size": {
|
| 147 |
+
"longest_edge": 384
|
| 148 |
+
},
|
| 149 |
+
"model_type": "idefics3_vision",
|
| 150 |
+
"num_attention_heads": 16,
|
| 151 |
+
"num_channels": 3,
|
| 152 |
+
"num_hidden_layers": 27,
|
| 153 |
+
"patch_size": 14,
|
| 154 |
+
"size": {
|
| 155 |
+
"longest_edge": 1920
|
| 156 |
+
},
|
| 157 |
+
"tie_word_embeddings": false
|
| 158 |
+
},
|
| 159 |
+
"vocab_size": 49155
|
| 160 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 0,
|
| 4 |
+
"eos_token_id": 49154,
|
| 5 |
+
"pad_token_id": 2,
|
| 6 |
+
"transformers_version": "4.57.0"
|
| 7 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8d787c8fc0b458a298166e4fe51db445a31962cf563613551f672d473c7150b0
|
| 3 |
+
size 4492630256
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d145c72bdffa85cea0e1d3a503d4388c6b85c83ac15a849977b766b7371e311c
|
| 3 |
+
size 5171520935
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:95d744506ed8242dbe82c0f3357716f73248e5153ff68604326958faa28d9296
|
| 3 |
+
size 14645
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bb53595bb041d4f6bfcf8bdb1611ba64ac44f2528fecf6fbcf3d23aba6e3e8dc
|
| 3 |
+
size 1465
|
trainer_state.json
ADDED
|
@@ -0,0 +1,643 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.7777777777777778,
|
| 6 |
+
"eval_steps": 50,
|
| 7 |
+
"global_step": 700,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0011111111111111111,
|
| 14 |
+
"grad_norm": 234.0,
|
| 15 |
+
"learning_rate": 0.0,
|
| 16 |
+
"loss": 34.6327,
|
| 17 |
+
"step": 1
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.011111111111111112,
|
| 21 |
+
"grad_norm": 196.0,
|
| 22 |
+
"learning_rate": 9e-06,
|
| 23 |
+
"loss": 32.4195,
|
| 24 |
+
"step": 10
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.022222222222222223,
|
| 28 |
+
"grad_norm": 66.875,
|
| 29 |
+
"learning_rate": 1.9e-05,
|
| 30 |
+
"loss": 14.1658,
|
| 31 |
+
"step": 20
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.03333333333333333,
|
| 35 |
+
"grad_norm": 0.9267578125,
|
| 36 |
+
"learning_rate": 2.9e-05,
|
| 37 |
+
"loss": 0.4885,
|
| 38 |
+
"step": 30
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.044444444444444446,
|
| 42 |
+
"grad_norm": 0.4169921875,
|
| 43 |
+
"learning_rate": 3.9000000000000006e-05,
|
| 44 |
+
"loss": 0.3108,
|
| 45 |
+
"step": 40
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.05555555555555555,
|
| 49 |
+
"grad_norm": 0.41357421875,
|
| 50 |
+
"learning_rate": 4.9e-05,
|
| 51 |
+
"loss": 0.2811,
|
| 52 |
+
"step": 50
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.05555555555555555,
|
| 56 |
+
"eval_loss": 0.14292466640472412,
|
| 57 |
+
"eval_runtime": 449.904,
|
| 58 |
+
"eval_samples_per_second": 4.001,
|
| 59 |
+
"eval_steps_per_second": 2.0,
|
| 60 |
+
"step": 50
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.06666666666666667,
|
| 64 |
+
"grad_norm": 0.359619140625,
|
| 65 |
+
"learning_rate": 4.9742857142857145e-05,
|
| 66 |
+
"loss": 0.2584,
|
| 67 |
+
"step": 60
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.07777777777777778,
|
| 71 |
+
"grad_norm": 0.431640625,
|
| 72 |
+
"learning_rate": 4.9457142857142854e-05,
|
| 73 |
+
"loss": 0.2643,
|
| 74 |
+
"step": 70
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.08888888888888889,
|
| 78 |
+
"grad_norm": 0.46923828125,
|
| 79 |
+
"learning_rate": 4.917142857142858e-05,
|
| 80 |
+
"loss": 0.2923,
|
| 81 |
+
"step": 80
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.1,
|
| 85 |
+
"grad_norm": 0.449951171875,
|
| 86 |
+
"learning_rate": 4.888571428571429e-05,
|
| 87 |
+
"loss": 0.29,
|
| 88 |
+
"step": 90
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 0.1111111111111111,
|
| 92 |
+
"grad_norm": 0.3896484375,
|
| 93 |
+
"learning_rate": 4.86e-05,
|
| 94 |
+
"loss": 0.2609,
|
| 95 |
+
"step": 100
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 0.1111111111111111,
|
| 99 |
+
"eval_loss": 0.13491107523441315,
|
| 100 |
+
"eval_runtime": 431.8863,
|
| 101 |
+
"eval_samples_per_second": 4.168,
|
| 102 |
+
"eval_steps_per_second": 2.084,
|
| 103 |
+
"step": 100
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.12222222222222222,
|
| 107 |
+
"grad_norm": 0.3828125,
|
| 108 |
+
"learning_rate": 4.831428571428572e-05,
|
| 109 |
+
"loss": 0.2863,
|
| 110 |
+
"step": 110
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.13333333333333333,
|
| 114 |
+
"grad_norm": 0.603515625,
|
| 115 |
+
"learning_rate": 4.802857142857143e-05,
|
| 116 |
+
"loss": 0.2647,
|
| 117 |
+
"step": 120
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.14444444444444443,
|
| 121 |
+
"grad_norm": 0.439208984375,
|
| 122 |
+
"learning_rate": 4.7742857142857144e-05,
|
| 123 |
+
"loss": 0.2706,
|
| 124 |
+
"step": 130
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 0.15555555555555556,
|
| 128 |
+
"grad_norm": 0.50634765625,
|
| 129 |
+
"learning_rate": 4.745714285714286e-05,
|
| 130 |
+
"loss": 0.2356,
|
| 131 |
+
"step": 140
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 0.16666666666666666,
|
| 135 |
+
"grad_norm": 0.513671875,
|
| 136 |
+
"learning_rate": 4.717142857142857e-05,
|
| 137 |
+
"loss": 0.2469,
|
| 138 |
+
"step": 150
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.16666666666666666,
|
| 142 |
+
"eval_loss": 0.12212829291820526,
|
| 143 |
+
"eval_runtime": 424.7761,
|
| 144 |
+
"eval_samples_per_second": 4.238,
|
| 145 |
+
"eval_steps_per_second": 2.119,
|
| 146 |
+
"step": 150
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.17777777777777778,
|
| 150 |
+
"grad_norm": 0.48095703125,
|
| 151 |
+
"learning_rate": 4.6885714285714285e-05,
|
| 152 |
+
"loss": 0.2469,
|
| 153 |
+
"step": 160
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.18888888888888888,
|
| 157 |
+
"grad_norm": 0.4775390625,
|
| 158 |
+
"learning_rate": 4.660000000000001e-05,
|
| 159 |
+
"loss": 0.2566,
|
| 160 |
+
"step": 170
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.2,
|
| 164 |
+
"grad_norm": 0.416259765625,
|
| 165 |
+
"learning_rate": 4.631428571428572e-05,
|
| 166 |
+
"loss": 0.2263,
|
| 167 |
+
"step": 180
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 0.2111111111111111,
|
| 171 |
+
"grad_norm": 0.47900390625,
|
| 172 |
+
"learning_rate": 4.602857142857143e-05,
|
| 173 |
+
"loss": 0.2505,
|
| 174 |
+
"step": 190
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.2222222222222222,
|
| 178 |
+
"grad_norm": 0.3916015625,
|
| 179 |
+
"learning_rate": 4.574285714285714e-05,
|
| 180 |
+
"loss": 0.2232,
|
| 181 |
+
"step": 200
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"epoch": 0.2222222222222222,
|
| 185 |
+
"eval_loss": 0.11928859353065491,
|
| 186 |
+
"eval_runtime": 432.6807,
|
| 187 |
+
"eval_samples_per_second": 4.16,
|
| 188 |
+
"eval_steps_per_second": 2.08,
|
| 189 |
+
"step": 200
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.23333333333333334,
|
| 193 |
+
"grad_norm": 0.48046875,
|
| 194 |
+
"learning_rate": 4.545714285714286e-05,
|
| 195 |
+
"loss": 0.2394,
|
| 196 |
+
"step": 210
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 0.24444444444444444,
|
| 200 |
+
"grad_norm": 0.40771484375,
|
| 201 |
+
"learning_rate": 4.5171428571428575e-05,
|
| 202 |
+
"loss": 0.226,
|
| 203 |
+
"step": 220
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 0.25555555555555554,
|
| 207 |
+
"grad_norm": 0.476318359375,
|
| 208 |
+
"learning_rate": 4.4885714285714284e-05,
|
| 209 |
+
"loss": 0.2628,
|
| 210 |
+
"step": 230
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 0.26666666666666666,
|
| 214 |
+
"grad_norm": 0.443603515625,
|
| 215 |
+
"learning_rate": 4.46e-05,
|
| 216 |
+
"loss": 0.25,
|
| 217 |
+
"step": 240
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 0.2777777777777778,
|
| 221 |
+
"grad_norm": 0.468994140625,
|
| 222 |
+
"learning_rate": 4.4314285714285716e-05,
|
| 223 |
+
"loss": 0.2663,
|
| 224 |
+
"step": 250
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.2777777777777778,
|
| 228 |
+
"eval_loss": 0.11743941903114319,
|
| 229 |
+
"eval_runtime": 463.571,
|
| 230 |
+
"eval_samples_per_second": 3.883,
|
| 231 |
+
"eval_steps_per_second": 1.941,
|
| 232 |
+
"step": 250
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 0.28888888888888886,
|
| 236 |
+
"grad_norm": 0.5185546875,
|
| 237 |
+
"learning_rate": 4.402857142857143e-05,
|
| 238 |
+
"loss": 0.2446,
|
| 239 |
+
"step": 260
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 0.3,
|
| 243 |
+
"grad_norm": 0.43994140625,
|
| 244 |
+
"learning_rate": 4.374285714285715e-05,
|
| 245 |
+
"loss": 0.257,
|
| 246 |
+
"step": 270
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 0.3111111111111111,
|
| 250 |
+
"grad_norm": 0.447021484375,
|
| 251 |
+
"learning_rate": 4.345714285714286e-05,
|
| 252 |
+
"loss": 0.2292,
|
| 253 |
+
"step": 280
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 0.32222222222222224,
|
| 257 |
+
"grad_norm": 0.372314453125,
|
| 258 |
+
"learning_rate": 4.317142857142857e-05,
|
| 259 |
+
"loss": 0.2314,
|
| 260 |
+
"step": 290
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.3333333333333333,
|
| 264 |
+
"grad_norm": 0.4365234375,
|
| 265 |
+
"learning_rate": 4.288571428571429e-05,
|
| 266 |
+
"loss": 0.2288,
|
| 267 |
+
"step": 300
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.3333333333333333,
|
| 271 |
+
"eval_loss": 0.11491398513317108,
|
| 272 |
+
"eval_runtime": 454.5263,
|
| 273 |
+
"eval_samples_per_second": 3.96,
|
| 274 |
+
"eval_steps_per_second": 1.98,
|
| 275 |
+
"step": 300
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.34444444444444444,
|
| 279 |
+
"grad_norm": 0.3798828125,
|
| 280 |
+
"learning_rate": 4.26e-05,
|
| 281 |
+
"loss": 0.23,
|
| 282 |
+
"step": 310
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.35555555555555557,
|
| 286 |
+
"grad_norm": 0.431396484375,
|
| 287 |
+
"learning_rate": 4.2314285714285715e-05,
|
| 288 |
+
"loss": 0.2208,
|
| 289 |
+
"step": 320
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.36666666666666664,
|
| 293 |
+
"grad_norm": 0.369384765625,
|
| 294 |
+
"learning_rate": 4.202857142857143e-05,
|
| 295 |
+
"loss": 0.213,
|
| 296 |
+
"step": 330
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.37777777777777777,
|
| 300 |
+
"grad_norm": 0.46826171875,
|
| 301 |
+
"learning_rate": 4.174285714285715e-05,
|
| 302 |
+
"loss": 0.2453,
|
| 303 |
+
"step": 340
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.3888888888888889,
|
| 307 |
+
"grad_norm": 0.485107421875,
|
| 308 |
+
"learning_rate": 4.145714285714286e-05,
|
| 309 |
+
"loss": 0.2454,
|
| 310 |
+
"step": 350
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.3888888888888889,
|
| 314 |
+
"eval_loss": 0.11370400339365005,
|
| 315 |
+
"eval_runtime": 429.4907,
|
| 316 |
+
"eval_samples_per_second": 4.191,
|
| 317 |
+
"eval_steps_per_second": 2.096,
|
| 318 |
+
"step": 350
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.4,
|
| 322 |
+
"grad_norm": 0.42919921875,
|
| 323 |
+
"learning_rate": 4.117142857142857e-05,
|
| 324 |
+
"loss": 0.245,
|
| 325 |
+
"step": 360
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.4111111111111111,
|
| 329 |
+
"grad_norm": 0.38427734375,
|
| 330 |
+
"learning_rate": 4.088571428571429e-05,
|
| 331 |
+
"loss": 0.2255,
|
| 332 |
+
"step": 370
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.4222222222222222,
|
| 336 |
+
"grad_norm": 0.295654296875,
|
| 337 |
+
"learning_rate": 4.0600000000000004e-05,
|
| 338 |
+
"loss": 0.2266,
|
| 339 |
+
"step": 380
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.43333333333333335,
|
| 343 |
+
"grad_norm": 0.437744140625,
|
| 344 |
+
"learning_rate": 4.0314285714285714e-05,
|
| 345 |
+
"loss": 0.2438,
|
| 346 |
+
"step": 390
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.4444444444444444,
|
| 350 |
+
"grad_norm": 0.4521484375,
|
| 351 |
+
"learning_rate": 4.002857142857143e-05,
|
| 352 |
+
"loss": 0.2135,
|
| 353 |
+
"step": 400
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.4444444444444444,
|
| 357 |
+
"eval_loss": 0.11205437779426575,
|
| 358 |
+
"eval_runtime": 431.8391,
|
| 359 |
+
"eval_samples_per_second": 4.168,
|
| 360 |
+
"eval_steps_per_second": 2.084,
|
| 361 |
+
"step": 400
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.45555555555555555,
|
| 365 |
+
"grad_norm": 0.33154296875,
|
| 366 |
+
"learning_rate": 3.9742857142857146e-05,
|
| 367 |
+
"loss": 0.2081,
|
| 368 |
+
"step": 410
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.4666666666666667,
|
| 372 |
+
"grad_norm": 0.35595703125,
|
| 373 |
+
"learning_rate": 3.945714285714286e-05,
|
| 374 |
+
"loss": 0.2369,
|
| 375 |
+
"step": 420
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.4777777777777778,
|
| 379 |
+
"grad_norm": 0.5166015625,
|
| 380 |
+
"learning_rate": 3.917142857142858e-05,
|
| 381 |
+
"loss": 0.2116,
|
| 382 |
+
"step": 430
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.4888888888888889,
|
| 386 |
+
"grad_norm": 0.403076171875,
|
| 387 |
+
"learning_rate": 3.888571428571429e-05,
|
| 388 |
+
"loss": 0.2159,
|
| 389 |
+
"step": 440
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.5,
|
| 393 |
+
"grad_norm": 0.349609375,
|
| 394 |
+
"learning_rate": 3.86e-05,
|
| 395 |
+
"loss": 0.208,
|
| 396 |
+
"step": 450
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.5,
|
| 400 |
+
"eval_loss": 0.11105828732252121,
|
| 401 |
+
"eval_runtime": 468.2112,
|
| 402 |
+
"eval_samples_per_second": 3.844,
|
| 403 |
+
"eval_steps_per_second": 1.922,
|
| 404 |
+
"step": 450
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 0.5111111111111111,
|
| 408 |
+
"grad_norm": 0.4482421875,
|
| 409 |
+
"learning_rate": 3.831428571428571e-05,
|
| 410 |
+
"loss": 0.2459,
|
| 411 |
+
"step": 460
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.5222222222222223,
|
| 415 |
+
"grad_norm": 0.365478515625,
|
| 416 |
+
"learning_rate": 3.802857142857143e-05,
|
| 417 |
+
"loss": 0.2239,
|
| 418 |
+
"step": 470
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.5333333333333333,
|
| 422 |
+
"grad_norm": 0.397216796875,
|
| 423 |
+
"learning_rate": 3.7742857142857145e-05,
|
| 424 |
+
"loss": 0.2396,
|
| 425 |
+
"step": 480
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 0.5444444444444444,
|
| 429 |
+
"grad_norm": 0.384765625,
|
| 430 |
+
"learning_rate": 3.745714285714286e-05,
|
| 431 |
+
"loss": 0.2263,
|
| 432 |
+
"step": 490
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 0.5555555555555556,
|
| 436 |
+
"grad_norm": 0.403564453125,
|
| 437 |
+
"learning_rate": 3.717142857142858e-05,
|
| 438 |
+
"loss": 0.2225,
|
| 439 |
+
"step": 500
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 0.5555555555555556,
|
| 443 |
+
"eval_loss": 0.11006490141153336,
|
| 444 |
+
"eval_runtime": 450.548,
|
| 445 |
+
"eval_samples_per_second": 3.995,
|
| 446 |
+
"eval_steps_per_second": 1.998,
|
| 447 |
+
"step": 500
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.5666666666666667,
|
| 451 |
+
"grad_norm": 0.4404296875,
|
| 452 |
+
"learning_rate": 3.688571428571429e-05,
|
| 453 |
+
"loss": 0.2238,
|
| 454 |
+
"step": 510
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 0.5777777777777777,
|
| 458 |
+
"grad_norm": 0.354248046875,
|
| 459 |
+
"learning_rate": 3.66e-05,
|
| 460 |
+
"loss": 0.2032,
|
| 461 |
+
"step": 520
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.5888888888888889,
|
| 465 |
+
"grad_norm": 0.344970703125,
|
| 466 |
+
"learning_rate": 3.631428571428572e-05,
|
| 467 |
+
"loss": 0.222,
|
| 468 |
+
"step": 530
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 0.6,
|
| 472 |
+
"grad_norm": 0.3740234375,
|
| 473 |
+
"learning_rate": 3.602857142857143e-05,
|
| 474 |
+
"loss": 0.2329,
|
| 475 |
+
"step": 540
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.6111111111111112,
|
| 479 |
+
"grad_norm": 0.3486328125,
|
| 480 |
+
"learning_rate": 3.574285714285714e-05,
|
| 481 |
+
"loss": 0.2188,
|
| 482 |
+
"step": 550
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.6111111111111112,
|
| 486 |
+
"eval_loss": 0.10909327119588852,
|
| 487 |
+
"eval_runtime": 435.9811,
|
| 488 |
+
"eval_samples_per_second": 4.129,
|
| 489 |
+
"eval_steps_per_second": 2.064,
|
| 490 |
+
"step": 550
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.6222222222222222,
|
| 494 |
+
"grad_norm": 0.425537109375,
|
| 495 |
+
"learning_rate": 3.545714285714286e-05,
|
| 496 |
+
"loss": 0.215,
|
| 497 |
+
"step": 560
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 0.6333333333333333,
|
| 501 |
+
"grad_norm": 0.41357421875,
|
| 502 |
+
"learning_rate": 3.517142857142857e-05,
|
| 503 |
+
"loss": 0.242,
|
| 504 |
+
"step": 570
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.6444444444444445,
|
| 508 |
+
"grad_norm": 0.482666015625,
|
| 509 |
+
"learning_rate": 3.488571428571429e-05,
|
| 510 |
+
"loss": 0.2034,
|
| 511 |
+
"step": 580
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 0.6555555555555556,
|
| 515 |
+
"grad_norm": 0.49658203125,
|
| 516 |
+
"learning_rate": 3.46e-05,
|
| 517 |
+
"loss": 0.2241,
|
| 518 |
+
"step": 590
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 0.6666666666666666,
|
| 522 |
+
"grad_norm": 0.42041015625,
|
| 523 |
+
"learning_rate": 3.431428571428572e-05,
|
| 524 |
+
"loss": 0.2477,
|
| 525 |
+
"step": 600
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.6666666666666666,
|
| 529 |
+
"eval_loss": 0.10835430771112442,
|
| 530 |
+
"eval_runtime": 451.0734,
|
| 531 |
+
"eval_samples_per_second": 3.99,
|
| 532 |
+
"eval_steps_per_second": 1.995,
|
| 533 |
+
"step": 600
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.6777777777777778,
|
| 537 |
+
"grad_norm": 0.372314453125,
|
| 538 |
+
"learning_rate": 3.402857142857143e-05,
|
| 539 |
+
"loss": 0.2293,
|
| 540 |
+
"step": 610
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 0.6888888888888889,
|
| 544 |
+
"grad_norm": 0.387939453125,
|
| 545 |
+
"learning_rate": 3.374285714285714e-05,
|
| 546 |
+
"loss": 0.246,
|
| 547 |
+
"step": 620
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 0.7,
|
| 551 |
+
"grad_norm": 0.426025390625,
|
| 552 |
+
"learning_rate": 3.345714285714286e-05,
|
| 553 |
+
"loss": 0.2,
|
| 554 |
+
"step": 630
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.7111111111111111,
|
| 558 |
+
"grad_norm": 0.380126953125,
|
| 559 |
+
"learning_rate": 3.3171428571428574e-05,
|
| 560 |
+
"loss": 0.2184,
|
| 561 |
+
"step": 640
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.7222222222222222,
|
| 565 |
+
"grad_norm": 0.393310546875,
|
| 566 |
+
"learning_rate": 3.2885714285714284e-05,
|
| 567 |
+
"loss": 0.2092,
|
| 568 |
+
"step": 650
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 0.7222222222222222,
|
| 572 |
+
"eval_loss": 0.10704999417066574,
|
| 573 |
+
"eval_runtime": 434.4458,
|
| 574 |
+
"eval_samples_per_second": 4.143,
|
| 575 |
+
"eval_steps_per_second": 2.072,
|
| 576 |
+
"step": 650
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.7333333333333333,
|
| 580 |
+
"grad_norm": 0.37841796875,
|
| 581 |
+
"learning_rate": 3.26e-05,
|
| 582 |
+
"loss": 0.2078,
|
| 583 |
+
"step": 660
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.7444444444444445,
|
| 587 |
+
"grad_norm": 0.400634765625,
|
| 588 |
+
"learning_rate": 3.2314285714285716e-05,
|
| 589 |
+
"loss": 0.2358,
|
| 590 |
+
"step": 670
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.7555555555555555,
|
| 594 |
+
"grad_norm": 0.38037109375,
|
| 595 |
+
"learning_rate": 3.202857142857143e-05,
|
| 596 |
+
"loss": 0.2226,
|
| 597 |
+
"step": 680
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.7666666666666667,
|
| 601 |
+
"grad_norm": 0.4248046875,
|
| 602 |
+
"learning_rate": 3.174285714285715e-05,
|
| 603 |
+
"loss": 0.2351,
|
| 604 |
+
"step": 690
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.7777777777777778,
|
| 608 |
+
"grad_norm": 0.51953125,
|
| 609 |
+
"learning_rate": 3.145714285714286e-05,
|
| 610 |
+
"loss": 0.2525,
|
| 611 |
+
"step": 700
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.7777777777777778,
|
| 615 |
+
"eval_loss": 0.10690909624099731,
|
| 616 |
+
"eval_runtime": 432.8965,
|
| 617 |
+
"eval_samples_per_second": 4.158,
|
| 618 |
+
"eval_steps_per_second": 2.079,
|
| 619 |
+
"step": 700
|
| 620 |
+
}
|
| 621 |
+
],
|
| 622 |
+
"logging_steps": 10,
|
| 623 |
+
"max_steps": 1800,
|
| 624 |
+
"num_input_tokens_seen": 0,
|
| 625 |
+
"num_train_epochs": 2,
|
| 626 |
+
"save_steps": 50,
|
| 627 |
+
"stateful_callbacks": {
|
| 628 |
+
"TrainerControl": {
|
| 629 |
+
"args": {
|
| 630 |
+
"should_epoch_stop": false,
|
| 631 |
+
"should_evaluate": false,
|
| 632 |
+
"should_log": false,
|
| 633 |
+
"should_save": true,
|
| 634 |
+
"should_training_stop": false
|
| 635 |
+
},
|
| 636 |
+
"attributes": {}
|
| 637 |
+
}
|
| 638 |
+
},
|
| 639 |
+
"total_flos": 1.2387832165981747e+17,
|
| 640 |
+
"train_batch_size": 4,
|
| 641 |
+
"trial_name": null,
|
| 642 |
+
"trial_params": null
|
| 643 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d596ce9cf12538cc209f87a3bcb25d6de593d17eb0a779831b1f716a21e281f9
|
| 3 |
+
size 5777
|