Commit ·
11cc438
1
Parent(s): 3df5a0c
Upload folder using huggingface_hub
Browse files- README.md +21 -0
- adapter_config.json +21 -0
- adapter_model.bin +3 -0
- special_tokens_map.json +24 -0
- tokenizer.json +0 -0
- tokenizer.model +3 -0
- tokenizer_config.json +34 -0
- train_args.json +123 -0
- training_args.bin +3 -0
- training_logs.json +1523 -0
README.md
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: peft
|
| 3 |
+
---
|
| 4 |
+
## Training procedure
|
| 5 |
+
|
| 6 |
+
|
| 7 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 8 |
+
- quant_method: bitsandbytes
|
| 9 |
+
- load_in_8bit: False
|
| 10 |
+
- load_in_4bit: True
|
| 11 |
+
- llm_int8_threshold: 6.0
|
| 12 |
+
- llm_int8_skip_modules: None
|
| 13 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 14 |
+
- llm_int8_has_fp16_weight: False
|
| 15 |
+
- bnb_4bit_quant_type: nf4
|
| 16 |
+
- bnb_4bit_use_double_quant: True
|
| 17 |
+
- bnb_4bit_compute_dtype: float16
|
| 18 |
+
### Framework versions
|
| 19 |
+
|
| 20 |
+
|
| 21 |
+
- PEFT 0.5.0
|
adapter_config.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"auto_mapping": null,
|
| 3 |
+
"base_model_name_or_path": "models/openllama-3b",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"fan_in_fan_out": false,
|
| 6 |
+
"inference_mode": true,
|
| 7 |
+
"init_lora_weights": true,
|
| 8 |
+
"layers_pattern": null,
|
| 9 |
+
"layers_to_transform": null,
|
| 10 |
+
"lora_alpha": 16,
|
| 11 |
+
"lora_dropout": 0.1,
|
| 12 |
+
"modules_to_save": null,
|
| 13 |
+
"peft_type": "LORA",
|
| 14 |
+
"r": 64,
|
| 15 |
+
"revision": null,
|
| 16 |
+
"target_modules": [
|
| 17 |
+
"q_proj",
|
| 18 |
+
"v_proj"
|
| 19 |
+
],
|
| 20 |
+
"task_type": "SEQ_CLS"
|
| 21 |
+
}
|
adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d22a7ebdf4da37149e511d7ba0a893ac0585a3aea3615a443f50f78a920e6b30
|
| 3 |
+
size 85258545
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "<s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": true,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "</s>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": true,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": "</s>",
|
| 17 |
+
"unk_token": {
|
| 18 |
+
"content": "<unk>",
|
| 19 |
+
"lstrip": false,
|
| 20 |
+
"normalized": true,
|
| 21 |
+
"rstrip": false,
|
| 22 |
+
"single_word": false
|
| 23 |
+
}
|
| 24 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ab1b681ec7fc02fed5edd3026687d7a692a918c4dd8e150ca2e3994a6229843b
|
| 3 |
+
size 534194
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"__type": "AddedToken",
|
| 4 |
+
"content": "<s>",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": true,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false
|
| 9 |
+
},
|
| 10 |
+
"clean_up_tokenization_spaces": false,
|
| 11 |
+
"eos_token": {
|
| 12 |
+
"__type": "AddedToken",
|
| 13 |
+
"content": "</s>",
|
| 14 |
+
"lstrip": false,
|
| 15 |
+
"normalized": true,
|
| 16 |
+
"rstrip": false,
|
| 17 |
+
"single_word": false
|
| 18 |
+
},
|
| 19 |
+
"legacy": null,
|
| 20 |
+
"model_max_length": 2048,
|
| 21 |
+
"pad_token": null,
|
| 22 |
+
"sp_model_kwargs": {},
|
| 23 |
+
"spaces_between_special_tokens": false,
|
| 24 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 25 |
+
"unk_token": {
|
| 26 |
+
"__type": "AddedToken",
|
| 27 |
+
"content": "<unk>",
|
| 28 |
+
"lstrip": false,
|
| 29 |
+
"normalized": true,
|
| 30 |
+
"rstrip": false,
|
| 31 |
+
"single_word": false
|
| 32 |
+
},
|
| 33 |
+
"use_default_system_prompt": true
|
| 34 |
+
}
|
train_args.json
ADDED
|
@@ -0,0 +1,123 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"initial_model_dir": "models/openllama-3b",
|
| 3 |
+
"distribution_id": "code_easy",
|
| 4 |
+
"date_trained": "15/10/2023 10:44:45",
|
| 5 |
+
"output_dir": "models/classify_lora/openllama-3b-code_easy",
|
| 6 |
+
"overwrite_output_dir": false,
|
| 7 |
+
"do_train": false,
|
| 8 |
+
"do_eval": true,
|
| 9 |
+
"do_predict": false,
|
| 10 |
+
"evaluation_strategy": "steps",
|
| 11 |
+
"prediction_loss_only": false,
|
| 12 |
+
"per_device_train_batch_size": 16,
|
| 13 |
+
"per_device_eval_batch_size": 32,
|
| 14 |
+
"per_gpu_train_batch_size": null,
|
| 15 |
+
"per_gpu_eval_batch_size": null,
|
| 16 |
+
"gradient_accumulation_steps": 1,
|
| 17 |
+
"eval_accumulation_steps": null,
|
| 18 |
+
"eval_delay": 0,
|
| 19 |
+
"learning_rate": 7.2e-05,
|
| 20 |
+
"weight_decay": 0.0,
|
| 21 |
+
"adam_beta1": 0.9,
|
| 22 |
+
"adam_beta2": 0.999,
|
| 23 |
+
"adam_epsilon": 1e-08,
|
| 24 |
+
"max_grad_norm": 0.3,
|
| 25 |
+
"num_train_epochs": 2,
|
| 26 |
+
"max_steps": 100,
|
| 27 |
+
"lr_scheduler_type": "constant",
|
| 28 |
+
"warmup_ratio": 0.03,
|
| 29 |
+
"warmup_steps": 0,
|
| 30 |
+
"log_level": "passive",
|
| 31 |
+
"log_level_replica": "warning",
|
| 32 |
+
"log_on_each_node": true,
|
| 33 |
+
"logging_dir": "models/classify_lora/openllama-3b-code_easy/runs/Oct15_10-25-52_compute-permanent-node-978",
|
| 34 |
+
"logging_strategy": "steps",
|
| 35 |
+
"logging_first_step": false,
|
| 36 |
+
"logging_steps": 1,
|
| 37 |
+
"logging_nan_inf_filter": true,
|
| 38 |
+
"save_strategy": "steps",
|
| 39 |
+
"save_steps": 25,
|
| 40 |
+
"save_total_limit": 0,
|
| 41 |
+
"save_safetensors": false,
|
| 42 |
+
"save_on_each_node": false,
|
| 43 |
+
"no_cuda": false,
|
| 44 |
+
"use_cpu": false,
|
| 45 |
+
"use_mps_device": false,
|
| 46 |
+
"seed": 42,
|
| 47 |
+
"data_seed": null,
|
| 48 |
+
"jit_mode_eval": false,
|
| 49 |
+
"use_ipex": false,
|
| 50 |
+
"bf16": false,
|
| 51 |
+
"fp16": false,
|
| 52 |
+
"fp16_opt_level": "O1",
|
| 53 |
+
"half_precision_backend": "auto",
|
| 54 |
+
"bf16_full_eval": false,
|
| 55 |
+
"fp16_full_eval": false,
|
| 56 |
+
"tf32": null,
|
| 57 |
+
"local_rank": 0,
|
| 58 |
+
"ddp_backend": null,
|
| 59 |
+
"tpu_num_cores": null,
|
| 60 |
+
"tpu_metrics_debug": false,
|
| 61 |
+
"debug": [],
|
| 62 |
+
"dataloader_drop_last": false,
|
| 63 |
+
"eval_steps": 25,
|
| 64 |
+
"dataloader_num_workers": 0,
|
| 65 |
+
"past_index": -1,
|
| 66 |
+
"run_name": "train|models-classify_lora-openllama-3b-code_easy",
|
| 67 |
+
"disable_tqdm": false,
|
| 68 |
+
"remove_unused_columns": false,
|
| 69 |
+
"label_names": null,
|
| 70 |
+
"load_best_model_at_end": false,
|
| 71 |
+
"metric_for_best_model": "eval_code_easy_score",
|
| 72 |
+
"greater_is_better": true,
|
| 73 |
+
"ignore_data_skip": false,
|
| 74 |
+
"sharded_ddp": [],
|
| 75 |
+
"fsdp": [],
|
| 76 |
+
"fsdp_min_num_params": 0,
|
| 77 |
+
"fsdp_config": {
|
| 78 |
+
"min_num_params": 0,
|
| 79 |
+
"xla": false,
|
| 80 |
+
"xla_fsdp_grad_ckpt": false
|
| 81 |
+
},
|
| 82 |
+
"fsdp_transformer_layer_cls_to_wrap": null,
|
| 83 |
+
"deepspeed": "configs/ds_zero_1.json",
|
| 84 |
+
"label_smoothing_factor": 0.0,
|
| 85 |
+
"optim": "paged_adamw_32bit",
|
| 86 |
+
"optim_args": null,
|
| 87 |
+
"adafactor": false,
|
| 88 |
+
"group_by_length": false,
|
| 89 |
+
"length_column_name": "length",
|
| 90 |
+
"report_to": [
|
| 91 |
+
"wandb"
|
| 92 |
+
],
|
| 93 |
+
"ddp_find_unused_parameters": false,
|
| 94 |
+
"ddp_bucket_cap_mb": null,
|
| 95 |
+
"ddp_broadcast_buffers": null,
|
| 96 |
+
"dataloader_pin_memory": true,
|
| 97 |
+
"skip_memory_metrics": true,
|
| 98 |
+
"use_legacy_prediction_loop": false,
|
| 99 |
+
"push_to_hub": false,
|
| 100 |
+
"resume_from_checkpoint": null,
|
| 101 |
+
"hub_model_id": null,
|
| 102 |
+
"hub_strategy": "every_save",
|
| 103 |
+
"hub_token": null,
|
| 104 |
+
"hub_private_repo": false,
|
| 105 |
+
"hub_always_push": false,
|
| 106 |
+
"gradient_checkpointing": false,
|
| 107 |
+
"include_inputs_for_metrics": false,
|
| 108 |
+
"fp16_backend": "auto",
|
| 109 |
+
"push_to_hub_model_id": null,
|
| 110 |
+
"push_to_hub_organization": null,
|
| 111 |
+
"push_to_hub_token": null,
|
| 112 |
+
"_n_gpu": 1,
|
| 113 |
+
"mp_parameters": "",
|
| 114 |
+
"auto_find_batch_size": false,
|
| 115 |
+
"full_determinism": false,
|
| 116 |
+
"torchdynamo": null,
|
| 117 |
+
"ray_scope": "last",
|
| 118 |
+
"ddp_timeout": 1800,
|
| 119 |
+
"torch_compile": false,
|
| 120 |
+
"torch_compile_backend": null,
|
| 121 |
+
"torch_compile_mode": null,
|
| 122 |
+
"dispatch_batches": null
|
| 123 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6fe3aae985974d071b1925a597fe1c5e41ef80c21082d534b5c1d1e5400d15ae
|
| 3 |
+
size 5115
|
training_logs.json
ADDED
|
@@ -0,0 +1,1523 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"loss": 0.7143,
|
| 4 |
+
"learning_rate": 7.2e-05,
|
| 5 |
+
"epoch": 0.02,
|
| 6 |
+
"step": 1
|
| 7 |
+
},
|
| 8 |
+
{
|
| 9 |
+
"loss": 0.6911,
|
| 10 |
+
"learning_rate": 7.2e-05,
|
| 11 |
+
"epoch": 0.04,
|
| 12 |
+
"step": 2
|
| 13 |
+
},
|
| 14 |
+
{
|
| 15 |
+
"loss": 0.6963,
|
| 16 |
+
"learning_rate": 7.2e-05,
|
| 17 |
+
"epoch": 0.05,
|
| 18 |
+
"step": 3
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"loss": 0.678,
|
| 22 |
+
"learning_rate": 7.2e-05,
|
| 23 |
+
"epoch": 0.07,
|
| 24 |
+
"step": 4
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"loss": 0.6763,
|
| 28 |
+
"learning_rate": 7.2e-05,
|
| 29 |
+
"epoch": 0.09,
|
| 30 |
+
"step": 5
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"loss": 0.6766,
|
| 34 |
+
"learning_rate": 7.2e-05,
|
| 35 |
+
"epoch": 0.11,
|
| 36 |
+
"step": 6
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"loss": 0.6475,
|
| 40 |
+
"learning_rate": 7.2e-05,
|
| 41 |
+
"epoch": 0.12,
|
| 42 |
+
"step": 7
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"loss": 0.6429,
|
| 46 |
+
"learning_rate": 7.2e-05,
|
| 47 |
+
"epoch": 0.14,
|
| 48 |
+
"step": 8
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"loss": 0.6151,
|
| 52 |
+
"learning_rate": 7.2e-05,
|
| 53 |
+
"epoch": 0.16,
|
| 54 |
+
"step": 9
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"loss": 0.6278,
|
| 58 |
+
"learning_rate": 7.2e-05,
|
| 59 |
+
"epoch": 0.18,
|
| 60 |
+
"step": 10
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"loss": 0.6226,
|
| 64 |
+
"learning_rate": 7.2e-05,
|
| 65 |
+
"epoch": 0.19,
|
| 66 |
+
"step": 11
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"loss": 0.6017,
|
| 70 |
+
"learning_rate": 7.2e-05,
|
| 71 |
+
"epoch": 0.21,
|
| 72 |
+
"step": 12
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"loss": 0.6445,
|
| 76 |
+
"learning_rate": 7.2e-05,
|
| 77 |
+
"epoch": 0.23,
|
| 78 |
+
"step": 13
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"loss": 0.5827,
|
| 82 |
+
"learning_rate": 7.2e-05,
|
| 83 |
+
"epoch": 0.25,
|
| 84 |
+
"step": 14
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"loss": 0.5437,
|
| 88 |
+
"learning_rate": 7.2e-05,
|
| 89 |
+
"epoch": 0.26,
|
| 90 |
+
"step": 15
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"loss": 0.5466,
|
| 94 |
+
"learning_rate": 7.2e-05,
|
| 95 |
+
"epoch": 0.28,
|
| 96 |
+
"step": 16
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"loss": 0.5452,
|
| 100 |
+
"learning_rate": 7.2e-05,
|
| 101 |
+
"epoch": 0.3,
|
| 102 |
+
"step": 17
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"loss": 0.5531,
|
| 106 |
+
"learning_rate": 7.2e-05,
|
| 107 |
+
"epoch": 0.32,
|
| 108 |
+
"step": 18
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"loss": 0.5615,
|
| 112 |
+
"learning_rate": 7.2e-05,
|
| 113 |
+
"epoch": 0.33,
|
| 114 |
+
"step": 19
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"loss": 0.5586,
|
| 118 |
+
"learning_rate": 7.2e-05,
|
| 119 |
+
"epoch": 0.35,
|
| 120 |
+
"step": 20
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"loss": 0.5116,
|
| 124 |
+
"learning_rate": 7.2e-05,
|
| 125 |
+
"epoch": 0.37,
|
| 126 |
+
"step": 21
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"loss": 0.4264,
|
| 130 |
+
"learning_rate": 7.2e-05,
|
| 131 |
+
"epoch": 0.39,
|
| 132 |
+
"step": 22
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"loss": 0.3866,
|
| 136 |
+
"learning_rate": 7.2e-05,
|
| 137 |
+
"epoch": 0.4,
|
| 138 |
+
"step": 23
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"loss": 0.5063,
|
| 142 |
+
"learning_rate": 7.2e-05,
|
| 143 |
+
"epoch": 0.42,
|
| 144 |
+
"step": 24
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"loss": 0.4082,
|
| 148 |
+
"learning_rate": 7.2e-05,
|
| 149 |
+
"epoch": 0.44,
|
| 150 |
+
"step": 25
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"eval_code_easy_loss": 0.4812281131744385,
|
| 154 |
+
"eval_code_easy_score": -0.14910461008548737,
|
| 155 |
+
"eval_code_easy_brier_score": 0.14910461008548737,
|
| 156 |
+
"eval_code_easy_average_probability": 0.6539512872695923,
|
| 157 |
+
"eval_code_easy_accuracy": 0.85,
|
| 158 |
+
"eval_code_easy_probabilities": [
|
| 159 |
+
0.7876701354980469,
|
| 160 |
+
0.7523844242095947,
|
| 161 |
+
0.6472867131233215,
|
| 162 |
+
0.7906888723373413,
|
| 163 |
+
0.8586155772209167,
|
| 164 |
+
0.6918468475341797,
|
| 165 |
+
0.4946542978286743,
|
| 166 |
+
0.578050434589386,
|
| 167 |
+
0.5266337990760803,
|
| 168 |
+
0.5107197761535645,
|
| 169 |
+
0.505793035030365,
|
| 170 |
+
0.45479127764701843,
|
| 171 |
+
0.4748489260673523,
|
| 172 |
+
0.5174643397331238,
|
| 173 |
+
0.4289233088493347,
|
| 174 |
+
0.5851827263832092,
|
| 175 |
+
0.8242431879043579,
|
| 176 |
+
0.9364657402038574,
|
| 177 |
+
0.5783424377441406,
|
| 178 |
+
0.48775342106819153,
|
| 179 |
+
0.4637766480445862,
|
| 180 |
+
0.8675320148468018,
|
| 181 |
+
0.7406874299049377,
|
| 182 |
+
0.7409279346466064,
|
| 183 |
+
0.7196134924888611,
|
| 184 |
+
0.6729203462600708,
|
| 185 |
+
0.7667352557182312,
|
| 186 |
+
0.7618107795715332,
|
| 187 |
+
0.5047318935394287,
|
| 188 |
+
0.8554519414901733,
|
| 189 |
+
0.48701024055480957,
|
| 190 |
+
0.5184937119483948,
|
| 191 |
+
0.26061391830444336,
|
| 192 |
+
0.6388080716133118,
|
| 193 |
+
0.557035505771637,
|
| 194 |
+
0.541324257850647,
|
| 195 |
+
0.7926197052001953,
|
| 196 |
+
0.9460112452507019,
|
| 197 |
+
0.548197865486145,
|
| 198 |
+
0.7201337814331055,
|
| 199 |
+
0.7818894982337952,
|
| 200 |
+
0.6181257367134094,
|
| 201 |
+
0.6234219670295715,
|
| 202 |
+
0.6032276749610901,
|
| 203 |
+
0.5777942538261414,
|
| 204 |
+
0.7809272408485413,
|
| 205 |
+
0.41414424777030945,
|
| 206 |
+
0.5810375809669495,
|
| 207 |
+
0.5093526244163513,
|
| 208 |
+
0.5954176783561707,
|
| 209 |
+
0.5138558745384216,
|
| 210 |
+
0.43075305223464966,
|
| 211 |
+
0.6626322865486145,
|
| 212 |
+
0.6255874037742615,
|
| 213 |
+
0.7210639119148254,
|
| 214 |
+
0.9159994125366211,
|
| 215 |
+
0.8485010266304016,
|
| 216 |
+
0.8202032446861267,
|
| 217 |
+
0.5092943906784058,
|
| 218 |
+
0.8338488936424255,
|
| 219 |
+
0.5957618355751038,
|
| 220 |
+
0.6853424906730652,
|
| 221 |
+
0.6718733310699463,
|
| 222 |
+
0.9674582481384277,
|
| 223 |
+
0.9621556401252747,
|
| 224 |
+
0.9909061193466187,
|
| 225 |
+
0.5818647742271423,
|
| 226 |
+
0.5197274088859558,
|
| 227 |
+
0.5295705795288086,
|
| 228 |
+
0.5335188508033752,
|
| 229 |
+
0.35347408056259155,
|
| 230 |
+
0.20989884436130524,
|
| 231 |
+
0.8785598278045654,
|
| 232 |
+
0.9439873695373535,
|
| 233 |
+
0.8205457329750061,
|
| 234 |
+
0.5045193433761597,
|
| 235 |
+
0.5271017551422119,
|
| 236 |
+
0.7975461483001709,
|
| 237 |
+
0.5877244472503662,
|
| 238 |
+
0.5780063271522522,
|
| 239 |
+
0.4087151885032654,
|
| 240 |
+
0.502793550491333,
|
| 241 |
+
0.5167631506919861,
|
| 242 |
+
0.5050711035728455,
|
| 243 |
+
0.6043635010719299,
|
| 244 |
+
0.6559382677078247,
|
| 245 |
+
0.7178013920783997,
|
| 246 |
+
0.6761601567268372,
|
| 247 |
+
0.4603518843650818,
|
| 248 |
+
0.4613737463951111,
|
| 249 |
+
0.8092925548553467,
|
| 250 |
+
0.5021629929542542,
|
| 251 |
+
0.8891451358795166,
|
| 252 |
+
0.8492684960365295,
|
| 253 |
+
0.8078585863113403,
|
| 254 |
+
0.7930970191955566,
|
| 255 |
+
0.9665018916130066,
|
| 256 |
+
0.920259416103363,
|
| 257 |
+
0.9725727438926697,
|
| 258 |
+
0.6322199702262878
|
| 259 |
+
],
|
| 260 |
+
"eval_code_easy_runtime": 12.1001,
|
| 261 |
+
"eval_code_easy_samples_per_second": 8.264,
|
| 262 |
+
"eval_code_easy_steps_per_second": 0.165,
|
| 263 |
+
"epoch": 0.44,
|
| 264 |
+
"step": 25
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"eval_code_hard_loss": 0.6784273386001587,
|
| 268 |
+
"eval_code_hard_score": -0.23437322676181793,
|
| 269 |
+
"eval_code_hard_brier_score": 0.23437322676181793,
|
| 270 |
+
"eval_code_hard_average_probability": 0.5313071608543396,
|
| 271 |
+
"eval_code_hard_accuracy": 0.63,
|
| 272 |
+
"eval_code_hard_probabilities": [
|
| 273 |
+
0.6153736114501953,
|
| 274 |
+
0.5013247132301331,
|
| 275 |
+
0.5044454336166382,
|
| 276 |
+
0.9211879968643188,
|
| 277 |
+
0.8884113430976868,
|
| 278 |
+
0.9285586476325989,
|
| 279 |
+
0.4261794984340668,
|
| 280 |
+
0.4360635578632355,
|
| 281 |
+
0.5060375928878784,
|
| 282 |
+
0.4633253514766693,
|
| 283 |
+
0.49218645691871643,
|
| 284 |
+
0.5567600727081299,
|
| 285 |
+
0.878082811832428,
|
| 286 |
+
0.8190033435821533,
|
| 287 |
+
0.8284511566162109,
|
| 288 |
+
0.5461653470993042,
|
| 289 |
+
0.5189316868782043,
|
| 290 |
+
0.535224974155426,
|
| 291 |
+
0.42352545261383057,
|
| 292 |
+
0.37337490916252136,
|
| 293 |
+
0.3578466475009918,
|
| 294 |
+
0.5396072268486023,
|
| 295 |
+
0.5384335517883301,
|
| 296 |
+
0.5416436195373535,
|
| 297 |
+
0.4356071949005127,
|
| 298 |
+
0.5550840497016907,
|
| 299 |
+
0.38927942514419556,
|
| 300 |
+
0.19190631806850433,
|
| 301 |
+
0.525556743144989,
|
| 302 |
+
0.563071072101593,
|
| 303 |
+
0.5183289647102356,
|
| 304 |
+
0.5080891847610474,
|
| 305 |
+
0.4980936646461487,
|
| 306 |
+
0.4459080696105957,
|
| 307 |
+
0.4500540792942047,
|
| 308 |
+
0.44474658370018005,
|
| 309 |
+
0.35595303773880005,
|
| 310 |
+
0.3355133831501007,
|
| 311 |
+
0.7131393551826477,
|
| 312 |
+
0.5476568937301636,
|
| 313 |
+
0.6050811409950256,
|
| 314 |
+
0.6322606801986694,
|
| 315 |
+
0.5295588374137878,
|
| 316 |
+
0.5178883075714111,
|
| 317 |
+
0.5216538906097412,
|
| 318 |
+
0.6891106367111206,
|
| 319 |
+
0.4794466197490692,
|
| 320 |
+
0.5025456547737122,
|
| 321 |
+
0.7308195233345032,
|
| 322 |
+
0.37424957752227783,
|
| 323 |
+
0.37053611874580383,
|
| 324 |
+
0.5303871631622314,
|
| 325 |
+
0.5796085000038147,
|
| 326 |
+
0.7150017619132996,
|
| 327 |
+
0.4277779161930084,
|
| 328 |
+
0.4290866553783417,
|
| 329 |
+
0.430403470993042,
|
| 330 |
+
0.5324957370758057,
|
| 331 |
+
0.5010586380958557,
|
| 332 |
+
0.520031750202179,
|
| 333 |
+
0.6008161902427673,
|
| 334 |
+
0.5889564156532288,
|
| 335 |
+
0.5865479111671448,
|
| 336 |
+
0.5397530794143677,
|
| 337 |
+
0.5804182887077332,
|
| 338 |
+
0.44839367270469666,
|
| 339 |
+
0.5159912109375,
|
| 340 |
+
0.7570854425430298,
|
| 341 |
+
0.49450379610061646,
|
| 342 |
+
0.6515185236930847,
|
| 343 |
+
0.6266943216323853,
|
| 344 |
+
0.4965268075466156,
|
| 345 |
+
0.5120733976364136,
|
| 346 |
+
0.540576696395874,
|
| 347 |
+
0.5366629958152771,
|
| 348 |
+
0.4482656717300415,
|
| 349 |
+
0.46240511536598206,
|
| 350 |
+
0.5012778639793396,
|
| 351 |
+
0.4999728500843048,
|
| 352 |
+
0.4921656548976898,
|
| 353 |
+
0.5284487009048462,
|
| 354 |
+
0.5068923234939575,
|
| 355 |
+
0.46083176136016846,
|
| 356 |
+
0.5137518644332886,
|
| 357 |
+
0.5790409445762634,
|
| 358 |
+
0.515843391418457,
|
| 359 |
+
0.542677640914917,
|
| 360 |
+
0.5102853178977966,
|
| 361 |
+
0.4863445460796356,
|
| 362 |
+
0.49948781728744507,
|
| 363 |
+
0.5162002444267273,
|
| 364 |
+
0.5039041042327881,
|
| 365 |
+
0.4971469044685364,
|
| 366 |
+
0.4223201274871826,
|
| 367 |
+
0.4179053008556366,
|
| 368 |
+
0.43632903695106506,
|
| 369 |
+
0.5268281102180481,
|
| 370 |
+
0.5249642729759216,
|
| 371 |
+
0.4914223849773407,
|
| 372 |
+
0.532348096370697
|
| 373 |
+
],
|
| 374 |
+
"eval_code_hard_runtime": 53.2874,
|
| 375 |
+
"eval_code_hard_samples_per_second": 1.877,
|
| 376 |
+
"eval_code_hard_steps_per_second": 0.038,
|
| 377 |
+
"epoch": 0.44,
|
| 378 |
+
"step": 25
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"loss": 0.4493,
|
| 382 |
+
"learning_rate": 7.2e-05,
|
| 383 |
+
"epoch": 0.46,
|
| 384 |
+
"step": 26
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"loss": 0.4695,
|
| 388 |
+
"learning_rate": 7.2e-05,
|
| 389 |
+
"epoch": 0.47,
|
| 390 |
+
"step": 27
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"loss": 0.3858,
|
| 394 |
+
"learning_rate": 7.2e-05,
|
| 395 |
+
"epoch": 0.49,
|
| 396 |
+
"step": 28
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"loss": 0.3683,
|
| 400 |
+
"learning_rate": 7.2e-05,
|
| 401 |
+
"epoch": 0.51,
|
| 402 |
+
"step": 29
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"loss": 0.4011,
|
| 406 |
+
"learning_rate": 7.2e-05,
|
| 407 |
+
"epoch": 0.53,
|
| 408 |
+
"step": 30
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"loss": 0.346,
|
| 412 |
+
"learning_rate": 7.2e-05,
|
| 413 |
+
"epoch": 0.54,
|
| 414 |
+
"step": 31
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"loss": 0.2892,
|
| 418 |
+
"learning_rate": 7.2e-05,
|
| 419 |
+
"epoch": 0.56,
|
| 420 |
+
"step": 32
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"loss": 0.4798,
|
| 424 |
+
"learning_rate": 7.2e-05,
|
| 425 |
+
"epoch": 0.58,
|
| 426 |
+
"step": 33
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"loss": 0.3073,
|
| 430 |
+
"learning_rate": 7.2e-05,
|
| 431 |
+
"epoch": 0.6,
|
| 432 |
+
"step": 34
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"loss": 0.3376,
|
| 436 |
+
"learning_rate": 7.2e-05,
|
| 437 |
+
"epoch": 0.61,
|
| 438 |
+
"step": 35
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"loss": 0.2565,
|
| 442 |
+
"learning_rate": 7.2e-05,
|
| 443 |
+
"epoch": 0.63,
|
| 444 |
+
"step": 36
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"loss": 0.1994,
|
| 448 |
+
"learning_rate": 7.2e-05,
|
| 449 |
+
"epoch": 0.65,
|
| 450 |
+
"step": 37
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"loss": 0.3296,
|
| 454 |
+
"learning_rate": 7.2e-05,
|
| 455 |
+
"epoch": 0.67,
|
| 456 |
+
"step": 38
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"loss": 0.2029,
|
| 460 |
+
"learning_rate": 7.2e-05,
|
| 461 |
+
"epoch": 0.68,
|
| 462 |
+
"step": 39
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"loss": 0.2739,
|
| 466 |
+
"learning_rate": 7.2e-05,
|
| 467 |
+
"epoch": 0.7,
|
| 468 |
+
"step": 40
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"loss": 0.1792,
|
| 472 |
+
"learning_rate": 7.2e-05,
|
| 473 |
+
"epoch": 0.72,
|
| 474 |
+
"step": 41
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"loss": 0.1962,
|
| 478 |
+
"learning_rate": 7.2e-05,
|
| 479 |
+
"epoch": 0.74,
|
| 480 |
+
"step": 42
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"loss": 0.2281,
|
| 484 |
+
"learning_rate": 7.2e-05,
|
| 485 |
+
"epoch": 0.75,
|
| 486 |
+
"step": 43
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"loss": 0.1416,
|
| 490 |
+
"learning_rate": 7.2e-05,
|
| 491 |
+
"epoch": 0.77,
|
| 492 |
+
"step": 44
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"loss": 0.2981,
|
| 496 |
+
"learning_rate": 7.2e-05,
|
| 497 |
+
"epoch": 0.79,
|
| 498 |
+
"step": 45
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"loss": 0.2767,
|
| 502 |
+
"learning_rate": 7.2e-05,
|
| 503 |
+
"epoch": 0.81,
|
| 504 |
+
"step": 46
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"loss": 0.2888,
|
| 508 |
+
"learning_rate": 7.2e-05,
|
| 509 |
+
"epoch": 0.82,
|
| 510 |
+
"step": 47
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"loss": 0.3354,
|
| 514 |
+
"learning_rate": 7.2e-05,
|
| 515 |
+
"epoch": 0.84,
|
| 516 |
+
"step": 48
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"loss": 0.2017,
|
| 520 |
+
"learning_rate": 7.2e-05,
|
| 521 |
+
"epoch": 0.86,
|
| 522 |
+
"step": 49
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"loss": 0.1569,
|
| 526 |
+
"learning_rate": 7.2e-05,
|
| 527 |
+
"epoch": 0.88,
|
| 528 |
+
"step": 50
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"eval_code_easy_loss": 0.366973876953125,
|
| 532 |
+
"eval_code_easy_score": -0.11782870441675186,
|
| 533 |
+
"eval_code_easy_brier_score": 0.11782870441675186,
|
| 534 |
+
"eval_code_easy_average_probability": 0.7787162065505981,
|
| 535 |
+
"eval_code_easy_accuracy": 0.82,
|
| 536 |
+
"eval_code_easy_probabilities": [
|
| 537 |
+
0.9949461817741394,
|
| 538 |
+
0.9871755242347717,
|
| 539 |
+
0.9638071060180664,
|
| 540 |
+
0.9957897067070007,
|
| 541 |
+
0.9965255856513977,
|
| 542 |
+
0.9852939248085022,
|
| 543 |
+
0.6930150985717773,
|
| 544 |
+
0.9989522695541382,
|
| 545 |
+
0.8558874130249023,
|
| 546 |
+
0.4613070785999298,
|
| 547 |
+
0.510718822479248,
|
| 548 |
+
0.2324236035346985,
|
| 549 |
+
0.3973901569843292,
|
| 550 |
+
0.7853201031684875,
|
| 551 |
+
0.5421555042266846,
|
| 552 |
+
0.7354709506034851,
|
| 553 |
+
0.9784907698631287,
|
| 554 |
+
0.9970309734344482,
|
| 555 |
+
0.5461677312850952,
|
| 556 |
+
0.41464078426361084,
|
| 557 |
+
0.30067187547683716,
|
| 558 |
+
0.9925328493118286,
|
| 559 |
+
0.9674832224845886,
|
| 560 |
+
0.9843401908874512,
|
| 561 |
+
0.9914712309837341,
|
| 562 |
+
0.9825579524040222,
|
| 563 |
+
0.9267136454582214,
|
| 564 |
+
0.9984757304191589,
|
| 565 |
+
0.9873777031898499,
|
| 566 |
+
0.9998621940612793,
|
| 567 |
+
0.47989073395729065,
|
| 568 |
+
0.6140918731689453,
|
| 569 |
+
0.39963647723197937,
|
| 570 |
+
0.7302553653717041,
|
| 571 |
+
0.6465033292770386,
|
| 572 |
+
0.5506258606910706,
|
| 573 |
+
0.9697403311729431,
|
| 574 |
+
0.961764395236969,
|
| 575 |
+
0.4946594834327698,
|
| 576 |
+
0.8810425400733948,
|
| 577 |
+
0.8602923154830933,
|
| 578 |
+
0.7874042987823486,
|
| 579 |
+
0.9162008166313171,
|
| 580 |
+
0.9948663711547852,
|
| 581 |
+
0.9439269304275513,
|
| 582 |
+
0.9370085000991821,
|
| 583 |
+
0.34556975960731506,
|
| 584 |
+
0.10199112445116043,
|
| 585 |
+
0.6533010005950928,
|
| 586 |
+
0.9573982954025269,
|
| 587 |
+
0.7248471975326538,
|
| 588 |
+
0.2611381411552429,
|
| 589 |
+
0.9998348951339722,
|
| 590 |
+
0.9983940720558167,
|
| 591 |
+
0.9555704593658447,
|
| 592 |
+
0.999995231628418,
|
| 593 |
+
0.999896764755249,
|
| 594 |
+
0.9813823699951172,
|
| 595 |
+
0.5570517778396606,
|
| 596 |
+
0.9944754242897034,
|
| 597 |
+
0.6928525567054749,
|
| 598 |
+
0.9708505272865295,
|
| 599 |
+
0.9566367268562317,
|
| 600 |
+
0.999937891960144,
|
| 601 |
+
0.9999886751174927,
|
| 602 |
+
0.9999959468841553,
|
| 603 |
+
0.9741113781929016,
|
| 604 |
+
0.7828025221824646,
|
| 605 |
+
0.8869524002075195,
|
| 606 |
+
0.6031803488731384,
|
| 607 |
+
0.13089396059513092,
|
| 608 |
+
0.015088181011378765,
|
| 609 |
+
0.9989626407623291,
|
| 610 |
+
0.9993138313293457,
|
| 611 |
+
0.9666383862495422,
|
| 612 |
+
0.7539190649986267,
|
| 613 |
+
0.6800207495689392,
|
| 614 |
+
0.7904426455497742,
|
| 615 |
+
0.9921445250511169,
|
| 616 |
+
0.8796939849853516,
|
| 617 |
+
0.9624642133712769,
|
| 618 |
+
0.48169296979904175,
|
| 619 |
+
0.5490749478340149,
|
| 620 |
+
0.5619942545890808,
|
| 621 |
+
0.7675090432167053,
|
| 622 |
+
0.7946746349334717,
|
| 623 |
+
0.8194403648376465,
|
| 624 |
+
0.787605881690979,
|
| 625 |
+
0.2022314816713333,
|
| 626 |
+
0.3003339469432831,
|
| 627 |
+
0.999944806098938,
|
| 628 |
+
0.3906439542770386,
|
| 629 |
+
0.9998724460601807,
|
| 630 |
+
0.9990270137786865,
|
| 631 |
+
0.9992762207984924,
|
| 632 |
+
0.9981435537338257,
|
| 633 |
+
0.9994109869003296,
|
| 634 |
+
0.9930939674377441,
|
| 635 |
+
0.9995924830436707,
|
| 636 |
+
0.2884075939655304
|
| 637 |
+
],
|
| 638 |
+
"eval_code_easy_runtime": 12.0912,
|
| 639 |
+
"eval_code_easy_samples_per_second": 8.27,
|
| 640 |
+
"eval_code_easy_steps_per_second": 0.165,
|
| 641 |
+
"epoch": 0.88,
|
| 642 |
+
"step": 50
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"eval_code_hard_loss": 0.7648758888244629,
|
| 646 |
+
"eval_code_hard_score": -0.2455763965845108,
|
| 647 |
+
"eval_code_hard_brier_score": 0.2455763965845108,
|
| 648 |
+
"eval_code_hard_average_probability": 0.5492634773254395,
|
| 649 |
+
"eval_code_hard_accuracy": 0.58,
|
| 650 |
+
"eval_code_hard_probabilities": [
|
| 651 |
+
0.9692575335502625,
|
| 652 |
+
0.5152681469917297,
|
| 653 |
+
0.5391058921813965,
|
| 654 |
+
0.9931036829948425,
|
| 655 |
+
0.9909272193908691,
|
| 656 |
+
0.9973450303077698,
|
| 657 |
+
0.37631839513778687,
|
| 658 |
+
0.33356037735939026,
|
| 659 |
+
0.5611117482185364,
|
| 660 |
+
0.4919166564941406,
|
| 661 |
+
0.4896278977394104,
|
| 662 |
+
0.546688973903656,
|
| 663 |
+
0.9802632927894592,
|
| 664 |
+
0.9941868782043457,
|
| 665 |
+
0.9953175783157349,
|
| 666 |
+
0.5917196869850159,
|
| 667 |
+
0.556675374507904,
|
| 668 |
+
0.5831913352012634,
|
| 669 |
+
0.7293359041213989,
|
| 670 |
+
0.47196894884109497,
|
| 671 |
+
0.4489763081073761,
|
| 672 |
+
0.768726646900177,
|
| 673 |
+
0.5400140285491943,
|
| 674 |
+
0.6983548998832703,
|
| 675 |
+
0.38701897859573364,
|
| 676 |
+
0.6672759056091309,
|
| 677 |
+
0.38983210921287537,
|
| 678 |
+
0.005684987176209688,
|
| 679 |
+
0.6010024547576904,
|
| 680 |
+
0.6203393936157227,
|
| 681 |
+
0.590334415435791,
|
| 682 |
+
0.5500776767730713,
|
| 683 |
+
0.4848686158657074,
|
| 684 |
+
0.49481847882270813,
|
| 685 |
+
0.4460141360759735,
|
| 686 |
+
0.41459348797798157,
|
| 687 |
+
0.2456727772951126,
|
| 688 |
+
0.24025699496269226,
|
| 689 |
+
0.5689725875854492,
|
| 690 |
+
0.4383637607097626,
|
| 691 |
+
0.5424249172210693,
|
| 692 |
+
0.4941108524799347,
|
| 693 |
+
0.5239763855934143,
|
| 694 |
+
0.4846762716770172,
|
| 695 |
+
0.4753747582435608,
|
| 696 |
+
0.9676743149757385,
|
| 697 |
+
0.40851080417633057,
|
| 698 |
+
0.4941249489784241,
|
| 699 |
+
0.8557487726211548,
|
| 700 |
+
0.8221828937530518,
|
| 701 |
+
0.8013899922370911,
|
| 702 |
+
0.49702978134155273,
|
| 703 |
+
0.3039790391921997,
|
| 704 |
+
0.30129650235176086,
|
| 705 |
+
0.02055172249674797,
|
| 706 |
+
0.021210504695773125,
|
| 707 |
+
0.021720753982663155,
|
| 708 |
+
0.575912594795227,
|
| 709 |
+
0.5162267088890076,
|
| 710 |
+
0.5338138341903687,
|
| 711 |
+
0.8233920335769653,
|
| 712 |
+
0.8342075347900391,
|
| 713 |
+
0.8165725469589233,
|
| 714 |
+
0.6269811391830444,
|
| 715 |
+
0.6718840599060059,
|
| 716 |
+
0.5006963014602661,
|
| 717 |
+
0.3764708340167999,
|
| 718 |
+
0.48138299584388733,
|
| 719 |
+
0.33396115899086,
|
| 720 |
+
0.749821126461029,
|
| 721 |
+
0.6757022142410278,
|
| 722 |
+
0.48911863565444946,
|
| 723 |
+
0.5064474940299988,
|
| 724 |
+
0.62013179063797,
|
| 725 |
+
0.5993181467056274,
|
| 726 |
+
0.4782550632953644,
|
| 727 |
+
0.4940183758735657,
|
| 728 |
+
0.5602287650108337,
|
| 729 |
+
0.5086959004402161,
|
| 730 |
+
0.46642473340034485,
|
| 731 |
+
0.5594733953475952,
|
| 732 |
+
0.5183232426643372,
|
| 733 |
+
0.5857313275337219,
|
| 734 |
+
0.5656166076660156,
|
| 735 |
+
0.5952458381652832,
|
| 736 |
+
0.5976989269256592,
|
| 737 |
+
0.5781928300857544,
|
| 738 |
+
0.48465627431869507,
|
| 739 |
+
0.5290895700454712,
|
| 740 |
+
0.5022053718566895,
|
| 741 |
+
0.47955581545829773,
|
| 742 |
+
0.465986043214798,
|
| 743 |
+
0.46290040016174316,
|
| 744 |
+
0.33050066232681274,
|
| 745 |
+
0.34588003158569336,
|
| 746 |
+
0.387125700712204,
|
| 747 |
+
0.6185724139213562,
|
| 748 |
+
0.683625340461731,
|
| 749 |
+
0.32840457558631897,
|
| 750 |
+
0.723832905292511
|
| 751 |
+
],
|
| 752 |
+
"eval_code_hard_runtime": 53.3211,
|
| 753 |
+
"eval_code_hard_samples_per_second": 1.875,
|
| 754 |
+
"eval_code_hard_steps_per_second": 0.038,
|
| 755 |
+
"epoch": 0.88,
|
| 756 |
+
"step": 50
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"loss": 0.2551,
|
| 760 |
+
"learning_rate": 7.2e-05,
|
| 761 |
+
"epoch": 0.89,
|
| 762 |
+
"step": 51
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"loss": 0.1634,
|
| 766 |
+
"learning_rate": 7.2e-05,
|
| 767 |
+
"epoch": 0.91,
|
| 768 |
+
"step": 52
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"loss": 0.3866,
|
| 772 |
+
"learning_rate": 7.2e-05,
|
| 773 |
+
"epoch": 0.93,
|
| 774 |
+
"step": 53
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"loss": 0.4277,
|
| 778 |
+
"learning_rate": 7.2e-05,
|
| 779 |
+
"epoch": 0.95,
|
| 780 |
+
"step": 54
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"loss": 0.2225,
|
| 784 |
+
"learning_rate": 7.2e-05,
|
| 785 |
+
"epoch": 0.96,
|
| 786 |
+
"step": 55
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"loss": 0.1702,
|
| 790 |
+
"learning_rate": 7.2e-05,
|
| 791 |
+
"epoch": 0.98,
|
| 792 |
+
"step": 56
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"loss": 0.0772,
|
| 796 |
+
"learning_rate": 7.2e-05,
|
| 797 |
+
"epoch": 1.0,
|
| 798 |
+
"step": 57
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"loss": 0.082,
|
| 802 |
+
"learning_rate": 7.2e-05,
|
| 803 |
+
"epoch": 1.02,
|
| 804 |
+
"step": 58
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"loss": 0.3056,
|
| 808 |
+
"learning_rate": 7.2e-05,
|
| 809 |
+
"epoch": 1.04,
|
| 810 |
+
"step": 59
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"loss": 0.1856,
|
| 814 |
+
"learning_rate": 7.2e-05,
|
| 815 |
+
"epoch": 1.05,
|
| 816 |
+
"step": 60
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"loss": 0.1862,
|
| 820 |
+
"learning_rate": 7.2e-05,
|
| 821 |
+
"epoch": 1.07,
|
| 822 |
+
"step": 61
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"loss": 0.1939,
|
| 826 |
+
"learning_rate": 7.2e-05,
|
| 827 |
+
"epoch": 1.09,
|
| 828 |
+
"step": 62
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"loss": 0.1331,
|
| 832 |
+
"learning_rate": 7.2e-05,
|
| 833 |
+
"epoch": 1.11,
|
| 834 |
+
"step": 63
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"loss": 0.2338,
|
| 838 |
+
"learning_rate": 7.2e-05,
|
| 839 |
+
"epoch": 1.12,
|
| 840 |
+
"step": 64
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"loss": 0.1673,
|
| 844 |
+
"learning_rate": 7.2e-05,
|
| 845 |
+
"epoch": 1.14,
|
| 846 |
+
"step": 65
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"loss": 0.3696,
|
| 850 |
+
"learning_rate": 7.2e-05,
|
| 851 |
+
"epoch": 1.16,
|
| 852 |
+
"step": 66
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"loss": 0.1056,
|
| 856 |
+
"learning_rate": 7.2e-05,
|
| 857 |
+
"epoch": 1.18,
|
| 858 |
+
"step": 67
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"loss": 0.1094,
|
| 862 |
+
"learning_rate": 7.2e-05,
|
| 863 |
+
"epoch": 1.19,
|
| 864 |
+
"step": 68
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"loss": 0.0796,
|
| 868 |
+
"learning_rate": 7.2e-05,
|
| 869 |
+
"epoch": 1.21,
|
| 870 |
+
"step": 69
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"loss": 0.2261,
|
| 874 |
+
"learning_rate": 7.2e-05,
|
| 875 |
+
"epoch": 1.23,
|
| 876 |
+
"step": 70
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"loss": 0.1961,
|
| 880 |
+
"learning_rate": 7.2e-05,
|
| 881 |
+
"epoch": 1.25,
|
| 882 |
+
"step": 71
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"loss": 0.1452,
|
| 886 |
+
"learning_rate": 7.2e-05,
|
| 887 |
+
"epoch": 1.26,
|
| 888 |
+
"step": 72
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"loss": 0.1716,
|
| 892 |
+
"learning_rate": 7.2e-05,
|
| 893 |
+
"epoch": 1.28,
|
| 894 |
+
"step": 73
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"loss": 0.1421,
|
| 898 |
+
"learning_rate": 7.2e-05,
|
| 899 |
+
"epoch": 1.3,
|
| 900 |
+
"step": 74
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"loss": 0.3305,
|
| 904 |
+
"learning_rate": 7.2e-05,
|
| 905 |
+
"epoch": 1.32,
|
| 906 |
+
"step": 75
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"eval_code_easy_loss": 0.32117652893066406,
|
| 910 |
+
"eval_code_easy_score": -0.09214462339878082,
|
| 911 |
+
"eval_code_easy_brier_score": 0.09214462339878082,
|
| 912 |
+
"eval_code_easy_average_probability": 0.8216841220855713,
|
| 913 |
+
"eval_code_easy_accuracy": 0.88,
|
| 914 |
+
"eval_code_easy_probabilities": [
|
| 915 |
+
0.9994999170303345,
|
| 916 |
+
0.9912821054458618,
|
| 917 |
+
0.9577344059944153,
|
| 918 |
+
0.9980210065841675,
|
| 919 |
+
0.9973551034927368,
|
| 920 |
+
0.9871847033500671,
|
| 921 |
+
0.6745311617851257,
|
| 922 |
+
0.9999939203262329,
|
| 923 |
+
0.9304938912391663,
|
| 924 |
+
0.5086526870727539,
|
| 925 |
+
0.5627270340919495,
|
| 926 |
+
0.2480248361825943,
|
| 927 |
+
0.49823904037475586,
|
| 928 |
+
0.8804905414581299,
|
| 929 |
+
0.5820022821426392,
|
| 930 |
+
0.748561680316925,
|
| 931 |
+
0.9955660700798035,
|
| 932 |
+
0.9997114539146423,
|
| 933 |
+
0.8057821393013,
|
| 934 |
+
0.5374618768692017,
|
| 935 |
+
0.2833597958087921,
|
| 936 |
+
0.9967227578163147,
|
| 937 |
+
0.9863321781158447,
|
| 938 |
+
0.9948417544364929,
|
| 939 |
+
0.9996953010559082,
|
| 940 |
+
0.9988725781440735,
|
| 941 |
+
0.9955350160598755,
|
| 942 |
+
0.9988835453987122,
|
| 943 |
+
0.9974863529205322,
|
| 944 |
+
0.9998946189880371,
|
| 945 |
+
0.8057632446289062,
|
| 946 |
+
0.7699967622756958,
|
| 947 |
+
0.6971409916877747,
|
| 948 |
+
0.7715607285499573,
|
| 949 |
+
0.6907534599304199,
|
| 950 |
+
0.5492137670516968,
|
| 951 |
+
0.9612501263618469,
|
| 952 |
+
0.9800459146499634,
|
| 953 |
+
0.6639916896820068,
|
| 954 |
+
0.992854654788971,
|
| 955 |
+
0.9864367842674255,
|
| 956 |
+
0.9040735363960266,
|
| 957 |
+
0.9259136319160461,
|
| 958 |
+
0.9870310425758362,
|
| 959 |
+
0.8883805274963379,
|
| 960 |
+
0.9945270419120789,
|
| 961 |
+
0.48652422428131104,
|
| 962 |
+
0.044553183019161224,
|
| 963 |
+
0.6629449129104614,
|
| 964 |
+
0.9678350687026978,
|
| 965 |
+
0.7551115155220032,
|
| 966 |
+
0.35034820437431335,
|
| 967 |
+
0.9999849796295166,
|
| 968 |
+
0.9999747276306152,
|
| 969 |
+
0.9967097043991089,
|
| 970 |
+
0.9999997615814209,
|
| 971 |
+
0.9999991655349731,
|
| 972 |
+
0.9840741157531738,
|
| 973 |
+
0.5545885562896729,
|
| 974 |
+
0.9983722567558289,
|
| 975 |
+
0.6852497458457947,
|
| 976 |
+
0.9241886138916016,
|
| 977 |
+
0.8779072165489197,
|
| 978 |
+
0.9999986886978149,
|
| 979 |
+
0.9999998807907104,
|
| 980 |
+
1.0,
|
| 981 |
+
0.9829393625259399,
|
| 982 |
+
0.9291936755180359,
|
| 983 |
+
0.9386154413223267,
|
| 984 |
+
0.6534070372581482,
|
| 985 |
+
0.10347169637680054,
|
| 986 |
+
0.0036570930387824774,
|
| 987 |
+
0.9999821186065674,
|
| 988 |
+
0.9999948740005493,
|
| 989 |
+
0.9956468939781189,
|
| 990 |
+
0.9652343988418579,
|
| 991 |
+
0.9181939363479614,
|
| 992 |
+
0.9902083873748779,
|
| 993 |
+
0.9989393353462219,
|
| 994 |
+
0.853695273399353,
|
| 995 |
+
0.9876751899719238,
|
| 996 |
+
0.4735547602176666,
|
| 997 |
+
0.5543529391288757,
|
| 998 |
+
0.5561679601669312,
|
| 999 |
+
0.8694782853126526,
|
| 1000 |
+
0.9327576756477356,
|
| 1001 |
+
0.9288151264190674,
|
| 1002 |
+
0.8425791263580322,
|
| 1003 |
+
0.2500445544719696,
|
| 1004 |
+
0.4530153274536133,
|
| 1005 |
+
0.9999988079071045,
|
| 1006 |
+
0.4624493420124054,
|
| 1007 |
+
0.9999938011169434,
|
| 1008 |
+
0.9999393224716187,
|
| 1009 |
+
0.9999752044677734,
|
| 1010 |
+
0.9998465776443481,
|
| 1011 |
+
0.999996542930603,
|
| 1012 |
+
0.9989469647407532,
|
| 1013 |
+
0.999995231628418,
|
| 1014 |
+
0.5134104490280151
|
| 1015 |
+
],
|
| 1016 |
+
"eval_code_easy_runtime": 12.0923,
|
| 1017 |
+
"eval_code_easy_samples_per_second": 8.27,
|
| 1018 |
+
"eval_code_easy_steps_per_second": 0.165,
|
| 1019 |
+
"epoch": 1.32,
|
| 1020 |
+
"step": 75
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"eval_code_hard_loss": 0.8514172434806824,
|
| 1024 |
+
"eval_code_hard_score": -0.2549550235271454,
|
| 1025 |
+
"eval_code_hard_brier_score": 0.2549550235271454,
|
| 1026 |
+
"eval_code_hard_average_probability": 0.5471080541610718,
|
| 1027 |
+
"eval_code_hard_accuracy": 0.63,
|
| 1028 |
+
"eval_code_hard_probabilities": [
|
| 1029 |
+
0.9908287525177002,
|
| 1030 |
+
0.5119895339012146,
|
| 1031 |
+
0.5428490042686462,
|
| 1032 |
+
0.9895175099372864,
|
| 1033 |
+
0.9891449809074402,
|
| 1034 |
+
0.9949464201927185,
|
| 1035 |
+
0.38589784502983093,
|
| 1036 |
+
0.2820265591144562,
|
| 1037 |
+
0.5328211784362793,
|
| 1038 |
+
0.5014193058013916,
|
| 1039 |
+
0.5086326003074646,
|
| 1040 |
+
0.5764115452766418,
|
| 1041 |
+
0.9859844446182251,
|
| 1042 |
+
0.9957466721534729,
|
| 1043 |
+
0.9967710375785828,
|
| 1044 |
+
0.5794450640678406,
|
| 1045 |
+
0.5391579270362854,
|
| 1046 |
+
0.6133323311805725,
|
| 1047 |
+
0.6606585383415222,
|
| 1048 |
+
0.34289708733558655,
|
| 1049 |
+
0.2988581657409668,
|
| 1050 |
+
0.788067102432251,
|
| 1051 |
+
0.542411744594574,
|
| 1052 |
+
0.7060100436210632,
|
| 1053 |
+
0.42542076110839844,
|
| 1054 |
+
0.7323933839797974,
|
| 1055 |
+
0.42784348130226135,
|
| 1056 |
+
0.0037993849255144596,
|
| 1057 |
+
0.6944400072097778,
|
| 1058 |
+
0.7083581686019897,
|
| 1059 |
+
0.5933657288551331,
|
| 1060 |
+
0.5459165573120117,
|
| 1061 |
+
0.5039231181144714,
|
| 1062 |
+
0.4072609841823578,
|
| 1063 |
+
0.3503766357898712,
|
| 1064 |
+
0.35972559452056885,
|
| 1065 |
+
0.1359257698059082,
|
| 1066 |
+
0.13460229337215424,
|
| 1067 |
+
0.4615738093852997,
|
| 1068 |
+
0.38321685791015625,
|
| 1069 |
+
0.37138813734054565,
|
| 1070 |
+
0.5392828583717346,
|
| 1071 |
+
0.5346622467041016,
|
| 1072 |
+
0.4690147042274475,
|
| 1073 |
+
0.4728128910064697,
|
| 1074 |
+
0.9811822772026062,
|
| 1075 |
+
0.36270952224731445,
|
| 1076 |
+
0.5393079519271851,
|
| 1077 |
+
0.8588457703590393,
|
| 1078 |
+
0.9537001252174377,
|
| 1079 |
+
0.9503171443939209,
|
| 1080 |
+
0.4992208480834961,
|
| 1081 |
+
0.16286799311637878,
|
| 1082 |
+
0.3865070939064026,
|
| 1083 |
+
0.0036291300784796476,
|
| 1084 |
+
0.0037814583629369736,
|
| 1085 |
+
0.003941776696592569,
|
| 1086 |
+
0.6453168392181396,
|
| 1087 |
+
0.5069921016693115,
|
| 1088 |
+
0.5409190654754639,
|
| 1089 |
+
0.620768666267395,
|
| 1090 |
+
0.6463491320610046,
|
| 1091 |
+
0.6067190766334534,
|
| 1092 |
+
0.6744793653488159,
|
| 1093 |
+
0.7356851696968079,
|
| 1094 |
+
0.6201030611991882,
|
| 1095 |
+
0.34589684009552,
|
| 1096 |
+
0.4320986270904541,
|
| 1097 |
+
0.2923872470855713,
|
| 1098 |
+
0.8295382857322693,
|
| 1099 |
+
0.7723568677902222,
|
| 1100 |
+
0.48901495337486267,
|
| 1101 |
+
0.5491488575935364,
|
| 1102 |
+
0.6467846035957336,
|
| 1103 |
+
0.6319132447242737,
|
| 1104 |
+
0.5377576351165771,
|
| 1105 |
+
0.5089011788368225,
|
| 1106 |
+
0.6081604957580566,
|
| 1107 |
+
0.5122814774513245,
|
| 1108 |
+
0.45855069160461426,
|
| 1109 |
+
0.5822516083717346,
|
| 1110 |
+
0.5393768548965454,
|
| 1111 |
+
0.579654335975647,
|
| 1112 |
+
0.5982511043548584,
|
| 1113 |
+
0.6504716873168945,
|
| 1114 |
+
0.658900797367096,
|
| 1115 |
+
0.6075130105018616,
|
| 1116 |
+
0.4958891272544861,
|
| 1117 |
+
0.5496628880500793,
|
| 1118 |
+
0.5090362429618835,
|
| 1119 |
+
0.46470561623573303,
|
| 1120 |
+
0.4267897605895996,
|
| 1121 |
+
0.43224194645881653,
|
| 1122 |
+
0.3274325728416443,
|
| 1123 |
+
0.35687658190727234,
|
| 1124 |
+
0.4236689805984497,
|
| 1125 |
+
0.6108702421188354,
|
| 1126 |
+
0.7473148107528687,
|
| 1127 |
+
0.3625641465187073,
|
| 1128 |
+
0.7580684423446655
|
| 1129 |
+
],
|
| 1130 |
+
"eval_code_hard_runtime": 53.2977,
|
| 1131 |
+
"eval_code_hard_samples_per_second": 1.876,
|
| 1132 |
+
"eval_code_hard_steps_per_second": 0.038,
|
| 1133 |
+
"epoch": 1.32,
|
| 1134 |
+
"step": 75
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"loss": 0.149,
|
| 1138 |
+
"learning_rate": 7.2e-05,
|
| 1139 |
+
"epoch": 1.33,
|
| 1140 |
+
"step": 76
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"loss": 0.2295,
|
| 1144 |
+
"learning_rate": 7.2e-05,
|
| 1145 |
+
"epoch": 1.35,
|
| 1146 |
+
"step": 77
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"loss": 0.1289,
|
| 1150 |
+
"learning_rate": 7.2e-05,
|
| 1151 |
+
"epoch": 1.37,
|
| 1152 |
+
"step": 78
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"loss": 0.0937,
|
| 1156 |
+
"learning_rate": 7.2e-05,
|
| 1157 |
+
"epoch": 1.39,
|
| 1158 |
+
"step": 79
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"loss": 0.2041,
|
| 1162 |
+
"learning_rate": 7.2e-05,
|
| 1163 |
+
"epoch": 1.4,
|
| 1164 |
+
"step": 80
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"loss": 0.2389,
|
| 1168 |
+
"learning_rate": 7.2e-05,
|
| 1169 |
+
"epoch": 1.42,
|
| 1170 |
+
"step": 81
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"loss": 0.1343,
|
| 1174 |
+
"learning_rate": 7.2e-05,
|
| 1175 |
+
"epoch": 1.44,
|
| 1176 |
+
"step": 82
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"loss": 0.0283,
|
| 1180 |
+
"learning_rate": 7.2e-05,
|
| 1181 |
+
"epoch": 1.46,
|
| 1182 |
+
"step": 83
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"loss": 0.0841,
|
| 1186 |
+
"learning_rate": 7.2e-05,
|
| 1187 |
+
"epoch": 1.47,
|
| 1188 |
+
"step": 84
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"loss": 0.3316,
|
| 1192 |
+
"learning_rate": 7.2e-05,
|
| 1193 |
+
"epoch": 1.49,
|
| 1194 |
+
"step": 85
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"loss": 0.0393,
|
| 1198 |
+
"learning_rate": 7.2e-05,
|
| 1199 |
+
"epoch": 1.51,
|
| 1200 |
+
"step": 86
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"loss": 0.0755,
|
| 1204 |
+
"learning_rate": 7.2e-05,
|
| 1205 |
+
"epoch": 1.53,
|
| 1206 |
+
"step": 87
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"loss": 0.1595,
|
| 1210 |
+
"learning_rate": 7.2e-05,
|
| 1211 |
+
"epoch": 1.54,
|
| 1212 |
+
"step": 88
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"loss": 0.1434,
|
| 1216 |
+
"learning_rate": 7.2e-05,
|
| 1217 |
+
"epoch": 1.56,
|
| 1218 |
+
"step": 89
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"loss": 0.1001,
|
| 1222 |
+
"learning_rate": 7.2e-05,
|
| 1223 |
+
"epoch": 1.58,
|
| 1224 |
+
"step": 90
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"loss": 0.1532,
|
| 1228 |
+
"learning_rate": 7.2e-05,
|
| 1229 |
+
"epoch": 1.6,
|
| 1230 |
+
"step": 91
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"loss": 0.1773,
|
| 1234 |
+
"learning_rate": 7.2e-05,
|
| 1235 |
+
"epoch": 1.61,
|
| 1236 |
+
"step": 92
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"loss": 0.1351,
|
| 1240 |
+
"learning_rate": 7.2e-05,
|
| 1241 |
+
"epoch": 1.63,
|
| 1242 |
+
"step": 93
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"loss": 0.0533,
|
| 1246 |
+
"learning_rate": 7.2e-05,
|
| 1247 |
+
"epoch": 1.65,
|
| 1248 |
+
"step": 94
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"loss": 0.1834,
|
| 1252 |
+
"learning_rate": 7.2e-05,
|
| 1253 |
+
"epoch": 1.67,
|
| 1254 |
+
"step": 95
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"loss": 0.1847,
|
| 1258 |
+
"learning_rate": 7.2e-05,
|
| 1259 |
+
"epoch": 1.68,
|
| 1260 |
+
"step": 96
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"loss": 0.072,
|
| 1264 |
+
"learning_rate": 7.2e-05,
|
| 1265 |
+
"epoch": 1.7,
|
| 1266 |
+
"step": 97
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"loss": 0.0431,
|
| 1270 |
+
"learning_rate": 7.2e-05,
|
| 1271 |
+
"epoch": 1.72,
|
| 1272 |
+
"step": 98
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"loss": 0.0796,
|
| 1276 |
+
"learning_rate": 7.2e-05,
|
| 1277 |
+
"epoch": 1.74,
|
| 1278 |
+
"step": 99
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"loss": 0.1992,
|
| 1282 |
+
"learning_rate": 7.2e-05,
|
| 1283 |
+
"epoch": 1.75,
|
| 1284 |
+
"step": 100
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"eval_code_easy_loss": 0.3075363039970398,
|
| 1288 |
+
"eval_code_easy_score": -0.07319054007530212,
|
| 1289 |
+
"eval_code_easy_brier_score": 0.07319054007530212,
|
| 1290 |
+
"eval_code_easy_average_probability": 0.8647841811180115,
|
| 1291 |
+
"eval_code_easy_accuracy": 0.92,
|
| 1292 |
+
"eval_code_easy_probabilities": [
|
| 1293 |
+
0.9999574422836304,
|
| 1294 |
+
0.9999752044677734,
|
| 1295 |
+
0.9896870255470276,
|
| 1296 |
+
0.9995444416999817,
|
| 1297 |
+
0.9996813535690308,
|
| 1298 |
+
0.9912283420562744,
|
| 1299 |
+
0.8707631230354309,
|
| 1300 |
+
1.0,
|
| 1301 |
+
0.9232298135757446,
|
| 1302 |
+
0.8876208066940308,
|
| 1303 |
+
0.7544382810592651,
|
| 1304 |
+
0.8350661993026733,
|
| 1305 |
+
0.8029730319976807,
|
| 1306 |
+
0.9684423804283142,
|
| 1307 |
+
0.8641867637634277,
|
| 1308 |
+
0.840520977973938,
|
| 1309 |
+
0.9997034668922424,
|
| 1310 |
+
0.9998855590820312,
|
| 1311 |
+
0.8682356476783752,
|
| 1312 |
+
0.5141866207122803,
|
| 1313 |
+
0.25204265117645264,
|
| 1314 |
+
0.9994714856147766,
|
| 1315 |
+
0.99544358253479,
|
| 1316 |
+
0.9964646100997925,
|
| 1317 |
+
0.9997562766075134,
|
| 1318 |
+
0.999864935874939,
|
| 1319 |
+
0.9721921682357788,
|
| 1320 |
+
0.9998952150344849,
|
| 1321 |
+
0.9750569462776184,
|
| 1322 |
+
0.9999854564666748,
|
| 1323 |
+
0.7429996132850647,
|
| 1324 |
+
0.822045624256134,
|
| 1325 |
+
0.7596055865287781,
|
| 1326 |
+
0.8471194505691528,
|
| 1327 |
+
0.7925592660903931,
|
| 1328 |
+
0.795616865158081,
|
| 1329 |
+
0.976042628288269,
|
| 1330 |
+
0.9971067309379578,
|
| 1331 |
+
0.5366230607032776,
|
| 1332 |
+
0.9954087138175964,
|
| 1333 |
+
0.977104127407074,
|
| 1334 |
+
0.9664822816848755,
|
| 1335 |
+
0.9948425889015198,
|
| 1336 |
+
0.9999990463256836,
|
| 1337 |
+
0.9974696636199951,
|
| 1338 |
+
0.9896783828735352,
|
| 1339 |
+
0.7055782675743103,
|
| 1340 |
+
0.010203487239778042,
|
| 1341 |
+
0.9344508051872253,
|
| 1342 |
+
0.999211311340332,
|
| 1343 |
+
0.9515018463134766,
|
| 1344 |
+
0.5253821611404419,
|
| 1345 |
+
1.0,
|
| 1346 |
+
0.9999994039535522,
|
| 1347 |
+
0.9999967813491821,
|
| 1348 |
+
1.0,
|
| 1349 |
+
1.0,
|
| 1350 |
+
0.9986414313316345,
|
| 1351 |
+
0.5329476594924927,
|
| 1352 |
+
0.9997962117195129,
|
| 1353 |
+
0.9566650986671448,
|
| 1354 |
+
0.9977226853370667,
|
| 1355 |
+
0.9972772002220154,
|
| 1356 |
+
1.0,
|
| 1357 |
+
1.0,
|
| 1358 |
+
1.0,
|
| 1359 |
+
0.9998082518577576,
|
| 1360 |
+
0.9984453320503235,
|
| 1361 |
+
0.9989882111549377,
|
| 1362 |
+
0.6647514700889587,
|
| 1363 |
+
0.06555651873350143,
|
| 1364 |
+
0.0002535696839913726,
|
| 1365 |
+
0.9999556541442871,
|
| 1366 |
+
0.9999914169311523,
|
| 1367 |
+
0.9952349066734314,
|
| 1368 |
+
0.9737288951873779,
|
| 1369 |
+
0.9164130091667175,
|
| 1370 |
+
0.9624624252319336,
|
| 1371 |
+
0.9998660087585449,
|
| 1372 |
+
0.9491363763809204,
|
| 1373 |
+
0.9998786449432373,
|
| 1374 |
+
0.4198010265827179,
|
| 1375 |
+
0.5238525867462158,
|
| 1376 |
+
0.5886799693107605,
|
| 1377 |
+
0.8423507213592529,
|
| 1378 |
+
0.9458127617835999,
|
| 1379 |
+
0.9036057591438293,
|
| 1380 |
+
0.9418210387229919,
|
| 1381 |
+
0.18787764012813568,
|
| 1382 |
+
0.38167285919189453,
|
| 1383 |
+
1.0,
|
| 1384 |
+
0.3317376375198364,
|
| 1385 |
+
0.9999998807907104,
|
| 1386 |
+
0.9999892711639404,
|
| 1387 |
+
0.9999970197677612,
|
| 1388 |
+
0.9999692440032959,
|
| 1389 |
+
0.9997808337211609,
|
| 1390 |
+
0.9896361231803894,
|
| 1391 |
+
0.9996919631958008,
|
| 1392 |
+
0.7980904579162598
|
| 1393 |
+
],
|
| 1394 |
+
"eval_code_easy_runtime": 12.0952,
|
| 1395 |
+
"eval_code_easy_samples_per_second": 8.268,
|
| 1396 |
+
"eval_code_easy_steps_per_second": 0.165,
|
| 1397 |
+
"epoch": 1.75,
|
| 1398 |
+
"step": 100
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"eval_code_hard_loss": 0.8713796734809875,
|
| 1402 |
+
"eval_code_hard_score": -0.23470951616764069,
|
| 1403 |
+
"eval_code_hard_brier_score": 0.23470951616764069,
|
| 1404 |
+
"eval_code_hard_average_probability": 0.5856627821922302,
|
| 1405 |
+
"eval_code_hard_accuracy": 0.65,
|
| 1406 |
+
"eval_code_hard_probabilities": [
|
| 1407 |
+
0.9855136275291443,
|
| 1408 |
+
0.49629148840904236,
|
| 1409 |
+
0.5387540459632874,
|
| 1410 |
+
0.9978917241096497,
|
| 1411 |
+
0.9978112578392029,
|
| 1412 |
+
0.9988699555397034,
|
| 1413 |
+
0.47865962982177734,
|
| 1414 |
+
0.3495783805847168,
|
| 1415 |
+
0.606380045413971,
|
| 1416 |
+
0.5135536789894104,
|
| 1417 |
+
0.5702087879180908,
|
| 1418 |
+
0.6042059659957886,
|
| 1419 |
+
0.9980809688568115,
|
| 1420 |
+
0.99955815076828,
|
| 1421 |
+
0.9996216297149658,
|
| 1422 |
+
0.5243353247642517,
|
| 1423 |
+
0.48392897844314575,
|
| 1424 |
+
0.6072529554367065,
|
| 1425 |
+
0.9629855751991272,
|
| 1426 |
+
0.8894397616386414,
|
| 1427 |
+
0.8946234583854675,
|
| 1428 |
+
0.8234318494796753,
|
| 1429 |
+
0.5752303004264832,
|
| 1430 |
+
0.7489052414894104,
|
| 1431 |
+
0.4938078820705414,
|
| 1432 |
+
0.764939546585083,
|
| 1433 |
+
0.49246543645858765,
|
| 1434 |
+
0.0005044332938268781,
|
| 1435 |
+
0.6989362835884094,
|
| 1436 |
+
0.6961206197738647,
|
| 1437 |
+
0.695743203163147,
|
| 1438 |
+
0.5966794490814209,
|
| 1439 |
+
0.4954136610031128,
|
| 1440 |
+
0.428554505109787,
|
| 1441 |
+
0.3025328516960144,
|
| 1442 |
+
0.31304794549942017,
|
| 1443 |
+
0.05157984420657158,
|
| 1444 |
+
0.057667434215545654,
|
| 1445 |
+
0.295099675655365,
|
| 1446 |
+
0.4117719233036041,
|
| 1447 |
+
0.4178787171840668,
|
| 1448 |
+
0.44591596722602844,
|
| 1449 |
+
0.5260471701622009,
|
| 1450 |
+
0.46641460061073303,
|
| 1451 |
+
0.44238483905792236,
|
| 1452 |
+
0.9921270608901978,
|
| 1453 |
+
0.3312618136405945,
|
| 1454 |
+
0.49680960178375244,
|
| 1455 |
+
0.8725456595420837,
|
| 1456 |
+
0.990997314453125,
|
| 1457 |
+
0.9901727437973022,
|
| 1458 |
+
0.42896220088005066,
|
| 1459 |
+
0.10010050982236862,
|
| 1460 |
+
0.30535706877708435,
|
| 1461 |
+
0.000553447927813977,
|
| 1462 |
+
0.000573858036659658,
|
| 1463 |
+
0.000650162051897496,
|
| 1464 |
+
0.6373094916343689,
|
| 1465 |
+
0.5249813199043274,
|
| 1466 |
+
0.5481424331665039,
|
| 1467 |
+
0.9771735668182373,
|
| 1468 |
+
0.9809019565582275,
|
| 1469 |
+
0.9766740798950195,
|
| 1470 |
+
0.7092797756195068,
|
| 1471 |
+
0.7087807059288025,
|
| 1472 |
+
0.700412392616272,
|
| 1473 |
+
0.4039199948310852,
|
| 1474 |
+
0.626448929309845,
|
| 1475 |
+
0.3165664076805115,
|
| 1476 |
+
0.8700190782546997,
|
| 1477 |
+
0.7991524934768677,
|
| 1478 |
+
0.5164951086044312,
|
| 1479 |
+
0.5391688346862793,
|
| 1480 |
+
0.6079928874969482,
|
| 1481 |
+
0.6740086674690247,
|
| 1482 |
+
0.5694684386253357,
|
| 1483 |
+
0.5098953247070312,
|
| 1484 |
+
0.6005474925041199,
|
| 1485 |
+
0.5051020979881287,
|
| 1486 |
+
0.4350145161151886,
|
| 1487 |
+
0.6045871376991272,
|
| 1488 |
+
0.5573720932006836,
|
| 1489 |
+
0.763244092464447,
|
| 1490 |
+
0.6566476225852966,
|
| 1491 |
+
0.655601978302002,
|
| 1492 |
+
0.6529899835586548,
|
| 1493 |
+
0.6917677521705627,
|
| 1494 |
+
0.4615822732448578,
|
| 1495 |
+
0.5680032968521118,
|
| 1496 |
+
0.5078142285346985,
|
| 1497 |
+
0.48479217290878296,
|
| 1498 |
+
0.437853068113327,
|
| 1499 |
+
0.4693388044834137,
|
| 1500 |
+
0.5179882645606995,
|
| 1501 |
+
0.5434749126434326,
|
| 1502 |
+
0.650637686252594,
|
| 1503 |
+
0.6607488393783569,
|
| 1504 |
+
0.7874378561973572,
|
| 1505 |
+
0.1389518529176712,
|
| 1506 |
+
0.76926189661026
|
| 1507 |
+
],
|
| 1508 |
+
"eval_code_hard_runtime": 53.3137,
|
| 1509 |
+
"eval_code_hard_samples_per_second": 1.876,
|
| 1510 |
+
"eval_code_hard_steps_per_second": 0.038,
|
| 1511 |
+
"epoch": 1.75,
|
| 1512 |
+
"step": 100
|
| 1513 |
+
},
|
| 1514 |
+
{
|
| 1515 |
+
"train_runtime": 1129.7886,
|
| 1516 |
+
"train_samples_per_second": 2.832,
|
| 1517 |
+
"train_steps_per_second": 0.089,
|
| 1518 |
+
"total_flos": 0.0,
|
| 1519 |
+
"train_loss": 0.30551302678883074,
|
| 1520 |
+
"epoch": 1.75,
|
| 1521 |
+
"step": 100
|
| 1522 |
+
}
|
| 1523 |
+
]
|