Commit ·
747a041
1
Parent(s): 07cf26b
Upload folder using huggingface_hub
Browse files- README.md +34 -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 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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 |
+
|
| 19 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 20 |
+
- quant_method: bitsandbytes
|
| 21 |
+
- load_in_8bit: False
|
| 22 |
+
- load_in_4bit: True
|
| 23 |
+
- llm_int8_threshold: 6.0
|
| 24 |
+
- llm_int8_skip_modules: None
|
| 25 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 26 |
+
- llm_int8_has_fp16_weight: False
|
| 27 |
+
- bnb_4bit_quant_type: nf4
|
| 28 |
+
- bnb_4bit_use_double_quant: True
|
| 29 |
+
- bnb_4bit_compute_dtype: float16
|
| 30 |
+
### Framework versions
|
| 31 |
+
|
| 32 |
+
- PEFT 0.5.0
|
| 33 |
+
|
| 34 |
+
- 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 sha256:8f7c32e1f9fbe3acd2c3f825f6a06d1663c87783781ca6d8f4f85f3bdb4857d2
|
| 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",
|
| 4 |
+
"date_trained": "15/10/2023 10:31:58",
|
| 5 |
+
"output_dir": "models/classify_lora/openllama-3b-code",
|
| 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/runs/Oct15_09-29-04_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",
|
| 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_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:cb98389d4837a0bbf000c2697be478ba29532b95bc9baf1835ecd3bac28bc353
|
| 3 |
+
size 5115
|
training_logs.json
ADDED
|
@@ -0,0 +1,1523 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"loss": 0.6836,
|
| 4 |
+
"learning_rate": 7.2e-05,
|
| 5 |
+
"epoch": 0.02,
|
| 6 |
+
"step": 1
|
| 7 |
+
},
|
| 8 |
+
{
|
| 9 |
+
"loss": 0.6947,
|
| 10 |
+
"learning_rate": 7.2e-05,
|
| 11 |
+
"epoch": 0.04,
|
| 12 |
+
"step": 2
|
| 13 |
+
},
|
| 14 |
+
{
|
| 15 |
+
"loss": 0.7068,
|
| 16 |
+
"learning_rate": 7.2e-05,
|
| 17 |
+
"epoch": 0.05,
|
| 18 |
+
"step": 3
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"loss": 0.6936,
|
| 22 |
+
"learning_rate": 7.2e-05,
|
| 23 |
+
"epoch": 0.07,
|
| 24 |
+
"step": 4
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"loss": 0.6894,
|
| 28 |
+
"learning_rate": 7.2e-05,
|
| 29 |
+
"epoch": 0.09,
|
| 30 |
+
"step": 5
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"loss": 0.6947,
|
| 34 |
+
"learning_rate": 7.2e-05,
|
| 35 |
+
"epoch": 0.11,
|
| 36 |
+
"step": 6
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"loss": 0.7127,
|
| 40 |
+
"learning_rate": 7.2e-05,
|
| 41 |
+
"epoch": 0.12,
|
| 42 |
+
"step": 7
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"loss": 0.6564,
|
| 46 |
+
"learning_rate": 7.2e-05,
|
| 47 |
+
"epoch": 0.14,
|
| 48 |
+
"step": 8
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"loss": 0.7051,
|
| 52 |
+
"learning_rate": 7.2e-05,
|
| 53 |
+
"epoch": 0.16,
|
| 54 |
+
"step": 9
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"loss": 0.6898,
|
| 58 |
+
"learning_rate": 7.2e-05,
|
| 59 |
+
"epoch": 0.18,
|
| 60 |
+
"step": 10
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"loss": 0.6269,
|
| 64 |
+
"learning_rate": 7.2e-05,
|
| 65 |
+
"epoch": 0.19,
|
| 66 |
+
"step": 11
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"loss": 0.6373,
|
| 70 |
+
"learning_rate": 7.2e-05,
|
| 71 |
+
"epoch": 0.21,
|
| 72 |
+
"step": 12
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"loss": 0.6626,
|
| 76 |
+
"learning_rate": 7.2e-05,
|
| 77 |
+
"epoch": 0.23,
|
| 78 |
+
"step": 13
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"loss": 0.671,
|
| 82 |
+
"learning_rate": 7.2e-05,
|
| 83 |
+
"epoch": 0.25,
|
| 84 |
+
"step": 14
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"loss": 0.6446,
|
| 88 |
+
"learning_rate": 7.2e-05,
|
| 89 |
+
"epoch": 0.26,
|
| 90 |
+
"step": 15
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"loss": 0.6511,
|
| 94 |
+
"learning_rate": 7.2e-05,
|
| 95 |
+
"epoch": 0.28,
|
| 96 |
+
"step": 16
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"loss": 0.6861,
|
| 100 |
+
"learning_rate": 7.2e-05,
|
| 101 |
+
"epoch": 0.3,
|
| 102 |
+
"step": 17
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"loss": 0.6569,
|
| 106 |
+
"learning_rate": 7.2e-05,
|
| 107 |
+
"epoch": 0.32,
|
| 108 |
+
"step": 18
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"loss": 0.6328,
|
| 112 |
+
"learning_rate": 7.2e-05,
|
| 113 |
+
"epoch": 0.33,
|
| 114 |
+
"step": 19
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"loss": 0.6302,
|
| 118 |
+
"learning_rate": 7.2e-05,
|
| 119 |
+
"epoch": 0.35,
|
| 120 |
+
"step": 20
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"loss": 0.6311,
|
| 124 |
+
"learning_rate": 7.2e-05,
|
| 125 |
+
"epoch": 0.37,
|
| 126 |
+
"step": 21
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"loss": 0.6667,
|
| 130 |
+
"learning_rate": 7.2e-05,
|
| 131 |
+
"epoch": 0.39,
|
| 132 |
+
"step": 22
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"loss": 0.6476,
|
| 136 |
+
"learning_rate": 7.2e-05,
|
| 137 |
+
"epoch": 0.4,
|
| 138 |
+
"step": 23
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"loss": 0.6099,
|
| 142 |
+
"learning_rate": 7.2e-05,
|
| 143 |
+
"epoch": 0.42,
|
| 144 |
+
"step": 24
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"loss": 0.636,
|
| 148 |
+
"learning_rate": 7.2e-05,
|
| 149 |
+
"epoch": 0.44,
|
| 150 |
+
"step": 25
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"eval_code_loss": 0.6500930190086365,
|
| 154 |
+
"eval_code_score": -0.23084141314029694,
|
| 155 |
+
"eval_code_brier_score": 0.23084141314029694,
|
| 156 |
+
"eval_code_average_probability": 0.5324862003326416,
|
| 157 |
+
"eval_code_accuracy": 0.63,
|
| 158 |
+
"eval_code_probabilities": [
|
| 159 |
+
0.8981996178627014,
|
| 160 |
+
0.9140064120292664,
|
| 161 |
+
0.9086233377456665,
|
| 162 |
+
0.4930489957332611,
|
| 163 |
+
0.5311785340309143,
|
| 164 |
+
0.49541160464286804,
|
| 165 |
+
0.5163961052894592,
|
| 166 |
+
0.5207482576370239,
|
| 167 |
+
0.4978530704975128,
|
| 168 |
+
0.4979601800441742,
|
| 169 |
+
0.5097731351852417,
|
| 170 |
+
0.5298001766204834,
|
| 171 |
+
0.4631284475326538,
|
| 172 |
+
0.4101221561431885,
|
| 173 |
+
0.3820548951625824,
|
| 174 |
+
0.5316497683525085,
|
| 175 |
+
0.4804396629333496,
|
| 176 |
+
0.5146058797836304,
|
| 177 |
+
0.5150214433670044,
|
| 178 |
+
0.4998961091041565,
|
| 179 |
+
0.5088847875595093,
|
| 180 |
+
0.5062065720558167,
|
| 181 |
+
0.49155697226524353,
|
| 182 |
+
0.5065352916717529,
|
| 183 |
+
0.3692309260368347,
|
| 184 |
+
0.3464958965778351,
|
| 185 |
+
0.345247745513916,
|
| 186 |
+
0.5082260370254517,
|
| 187 |
+
0.4985589385032654,
|
| 188 |
+
0.4908699691295624,
|
| 189 |
+
0.5014219284057617,
|
| 190 |
+
0.5078413486480713,
|
| 191 |
+
0.5118864178657532,
|
| 192 |
+
0.9250859618186951,
|
| 193 |
+
0.9192001819610596,
|
| 194 |
+
0.9183722734451294,
|
| 195 |
+
0.5567901134490967,
|
| 196 |
+
0.5061033368110657,
|
| 197 |
+
0.5202013850212097,
|
| 198 |
+
0.4856383204460144,
|
| 199 |
+
0.46390581130981445,
|
| 200 |
+
0.4597606658935547,
|
| 201 |
+
0.5073948502540588,
|
| 202 |
+
0.49284979701042175,
|
| 203 |
+
0.49324461817741394,
|
| 204 |
+
0.4991315007209778,
|
| 205 |
+
0.5046825408935547,
|
| 206 |
+
0.5030663013458252,
|
| 207 |
+
0.4927866756916046,
|
| 208 |
+
0.5034404993057251,
|
| 209 |
+
0.6324691772460938,
|
| 210 |
+
0.527839183807373,
|
| 211 |
+
0.6027267575263977,
|
| 212 |
+
0.5166205167770386,
|
| 213 |
+
0.5005240440368652,
|
| 214 |
+
0.5037304162979126,
|
| 215 |
+
0.5007357001304626,
|
| 216 |
+
0.5299914479255676,
|
| 217 |
+
0.49253302812576294,
|
| 218 |
+
0.5389838814735413,
|
| 219 |
+
0.7237929105758667,
|
| 220 |
+
0.722010612487793,
|
| 221 |
+
0.7248772978782654,
|
| 222 |
+
0.5261573791503906,
|
| 223 |
+
0.5009329319000244,
|
| 224 |
+
0.5066245198249817,
|
| 225 |
+
0.5175695419311523,
|
| 226 |
+
0.48236575722694397,
|
| 227 |
+
0.4813913404941559,
|
| 228 |
+
0.4951794743537903,
|
| 229 |
+
0.5201015472412109,
|
| 230 |
+
0.5013536810874939,
|
| 231 |
+
0.5012617111206055,
|
| 232 |
+
0.5066967606544495,
|
| 233 |
+
0.5075401067733765,
|
| 234 |
+
0.5807880759239197,
|
| 235 |
+
0.5756419897079468,
|
| 236 |
+
0.5374066829681396,
|
| 237 |
+
0.5401325821876526,
|
| 238 |
+
0.4964595437049866,
|
| 239 |
+
0.48420250415802,
|
| 240 |
+
0.5349816083908081,
|
| 241 |
+
0.4993498623371124,
|
| 242 |
+
0.5083227157592773,
|
| 243 |
+
0.4809667766094208,
|
| 244 |
+
0.5351281762123108,
|
| 245 |
+
0.45967379212379456,
|
| 246 |
+
0.5043829679489136,
|
| 247 |
+
0.4973975121974945,
|
| 248 |
+
0.5018096566200256,
|
| 249 |
+
0.4930133819580078,
|
| 250 |
+
0.5008613467216492,
|
| 251 |
+
0.5038902163505554,
|
| 252 |
+
0.5041508078575134,
|
| 253 |
+
0.4986605942249298,
|
| 254 |
+
0.5054576992988586,
|
| 255 |
+
0.4900255799293518,
|
| 256 |
+
0.4928497076034546,
|
| 257 |
+
0.5077279806137085,
|
| 258 |
+
0.49679625034332275
|
| 259 |
+
],
|
| 260 |
+
"eval_code_runtime": 36.4656,
|
| 261 |
+
"eval_code_samples_per_second": 2.742,
|
| 262 |
+
"eval_code_steps_per_second": 0.055,
|
| 263 |
+
"epoch": 0.44,
|
| 264 |
+
"step": 25
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"eval_counterfactual_python_loss": 0.6674560308456421,
|
| 268 |
+
"eval_counterfactual_python_score": -0.2368319183588028,
|
| 269 |
+
"eval_counterfactual_python_brier_score": 0.2368319183588028,
|
| 270 |
+
"eval_counterfactual_python_average_probability": 0.5182108879089355,
|
| 271 |
+
"eval_counterfactual_python_accuracy": 0.6,
|
| 272 |
+
"eval_counterfactual_python_probabilities": [
|
| 273 |
+
0.506840705871582,
|
| 274 |
+
0.48089414834976196,
|
| 275 |
+
0.497524231672287,
|
| 276 |
+
0.5044610500335693,
|
| 277 |
+
0.4899257719516754,
|
| 278 |
+
0.5029208660125732,
|
| 279 |
+
0.5528867840766907,
|
| 280 |
+
0.6020012497901917,
|
| 281 |
+
0.5903577208518982,
|
| 282 |
+
0.5012052059173584,
|
| 283 |
+
0.5010500550270081,
|
| 284 |
+
0.5014556050300598,
|
| 285 |
+
0.5129737257957458,
|
| 286 |
+
0.5166401863098145,
|
| 287 |
+
0.504987359046936,
|
| 288 |
+
0.48754236102104187,
|
| 289 |
+
0.4921533763408661,
|
| 290 |
+
0.5110666751861572,
|
| 291 |
+
0.5212812423706055,
|
| 292 |
+
0.505253791809082,
|
| 293 |
+
0.5044047236442566,
|
| 294 |
+
0.5446178913116455,
|
| 295 |
+
0.5504598021507263,
|
| 296 |
+
0.5181146860122681,
|
| 297 |
+
0.4997810125350952,
|
| 298 |
+
0.5025537610054016,
|
| 299 |
+
0.49983397126197815,
|
| 300 |
+
0.5592316389083862,
|
| 301 |
+
0.5561607480049133,
|
| 302 |
+
0.49701419472694397,
|
| 303 |
+
0.5067521929740906,
|
| 304 |
+
0.5109471082687378,
|
| 305 |
+
0.49662530422210693,
|
| 306 |
+
0.4926052391529083,
|
| 307 |
+
0.5500054955482483,
|
| 308 |
+
0.5147995352745056,
|
| 309 |
+
0.5712809562683105,
|
| 310 |
+
0.5422273874282837,
|
| 311 |
+
0.5272724032402039,
|
| 312 |
+
0.5182323455810547,
|
| 313 |
+
0.504339337348938,
|
| 314 |
+
0.524910569190979,
|
| 315 |
+
0.5017884373664856,
|
| 316 |
+
0.49858200550079346,
|
| 317 |
+
0.5003425478935242,
|
| 318 |
+
0.49639299511909485,
|
| 319 |
+
0.5009803175926208,
|
| 320 |
+
0.506009042263031,
|
| 321 |
+
0.5095840096473694,
|
| 322 |
+
0.49511101841926575,
|
| 323 |
+
0.49724477529525757,
|
| 324 |
+
0.5003291964530945,
|
| 325 |
+
0.49959462881088257,
|
| 326 |
+
0.5014533996582031,
|
| 327 |
+
0.5121566653251648,
|
| 328 |
+
0.4868036210536957,
|
| 329 |
+
0.4927564561367035,
|
| 330 |
+
0.49987566471099854,
|
| 331 |
+
0.48913922905921936,
|
| 332 |
+
0.4933817982673645,
|
| 333 |
+
0.5025197267532349,
|
| 334 |
+
0.4877781271934509,
|
| 335 |
+
0.4988517463207245,
|
| 336 |
+
0.48360976576805115,
|
| 337 |
+
0.47294488549232483,
|
| 338 |
+
0.44178009033203125,
|
| 339 |
+
0.5168694853782654,
|
| 340 |
+
0.5422465801239014,
|
| 341 |
+
0.5201556086540222,
|
| 342 |
+
0.5059149861335754,
|
| 343 |
+
0.49637845158576965,
|
| 344 |
+
0.518649697303772,
|
| 345 |
+
0.4840700328350067,
|
| 346 |
+
0.4878598749637604,
|
| 347 |
+
0.4397459328174591,
|
| 348 |
+
0.5004118084907532,
|
| 349 |
+
0.4982900023460388,
|
| 350 |
+
0.49845632910728455,
|
| 351 |
+
0.4866858422756195,
|
| 352 |
+
0.5425868630409241,
|
| 353 |
+
0.5578758120536804,
|
| 354 |
+
0.5089701414108276,
|
| 355 |
+
0.4936344027519226,
|
| 356 |
+
0.49497488141059875,
|
| 357 |
+
0.4820954203605652,
|
| 358 |
+
0.4383483827114105,
|
| 359 |
+
0.4935499131679535,
|
| 360 |
+
0.8830961585044861,
|
| 361 |
+
0.8793171048164368,
|
| 362 |
+
0.8719192743301392,
|
| 363 |
+
0.44867223501205444,
|
| 364 |
+
0.4727223813533783,
|
| 365 |
+
0.5310845971107483,
|
| 366 |
+
0.4810177981853485,
|
| 367 |
+
0.5083786249160767,
|
| 368 |
+
0.5173466205596924,
|
| 369 |
+
0.5266315340995789,
|
| 370 |
+
0.513174831867218,
|
| 371 |
+
0.5350879430770874,
|
| 372 |
+
0.5002694725990295
|
| 373 |
+
],
|
| 374 |
+
"eval_counterfactual_python_runtime": 42.2718,
|
| 375 |
+
"eval_counterfactual_python_samples_per_second": 2.366,
|
| 376 |
+
"eval_counterfactual_python_steps_per_second": 0.047,
|
| 377 |
+
"epoch": 0.44,
|
| 378 |
+
"step": 25
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"loss": 0.6523,
|
| 382 |
+
"learning_rate": 7.2e-05,
|
| 383 |
+
"epoch": 0.46,
|
| 384 |
+
"step": 26
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"loss": 0.6145,
|
| 388 |
+
"learning_rate": 7.2e-05,
|
| 389 |
+
"epoch": 0.47,
|
| 390 |
+
"step": 27
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"loss": 0.6282,
|
| 394 |
+
"learning_rate": 7.2e-05,
|
| 395 |
+
"epoch": 0.49,
|
| 396 |
+
"step": 28
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"loss": 0.5889,
|
| 400 |
+
"learning_rate": 7.2e-05,
|
| 401 |
+
"epoch": 0.51,
|
| 402 |
+
"step": 29
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"loss": 0.6173,
|
| 406 |
+
"learning_rate": 7.2e-05,
|
| 407 |
+
"epoch": 0.53,
|
| 408 |
+
"step": 30
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"loss": 0.6474,
|
| 412 |
+
"learning_rate": 7.2e-05,
|
| 413 |
+
"epoch": 0.54,
|
| 414 |
+
"step": 31
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"loss": 0.5995,
|
| 418 |
+
"learning_rate": 7.2e-05,
|
| 419 |
+
"epoch": 0.56,
|
| 420 |
+
"step": 32
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"loss": 0.5861,
|
| 424 |
+
"learning_rate": 7.2e-05,
|
| 425 |
+
"epoch": 0.58,
|
| 426 |
+
"step": 33
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"loss": 0.5736,
|
| 430 |
+
"learning_rate": 7.2e-05,
|
| 431 |
+
"epoch": 0.6,
|
| 432 |
+
"step": 34
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"loss": 0.6119,
|
| 436 |
+
"learning_rate": 7.2e-05,
|
| 437 |
+
"epoch": 0.61,
|
| 438 |
+
"step": 35
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"loss": 0.6247,
|
| 442 |
+
"learning_rate": 7.2e-05,
|
| 443 |
+
"epoch": 0.63,
|
| 444 |
+
"step": 36
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"loss": 0.6276,
|
| 448 |
+
"learning_rate": 7.2e-05,
|
| 449 |
+
"epoch": 0.65,
|
| 450 |
+
"step": 37
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"loss": 0.6098,
|
| 454 |
+
"learning_rate": 7.2e-05,
|
| 455 |
+
"epoch": 0.67,
|
| 456 |
+
"step": 38
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"loss": 0.6301,
|
| 460 |
+
"learning_rate": 7.2e-05,
|
| 461 |
+
"epoch": 0.68,
|
| 462 |
+
"step": 39
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"loss": 0.6021,
|
| 466 |
+
"learning_rate": 7.2e-05,
|
| 467 |
+
"epoch": 0.7,
|
| 468 |
+
"step": 40
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"loss": 0.555,
|
| 472 |
+
"learning_rate": 7.2e-05,
|
| 473 |
+
"epoch": 0.72,
|
| 474 |
+
"step": 41
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"loss": 0.5695,
|
| 478 |
+
"learning_rate": 7.2e-05,
|
| 479 |
+
"epoch": 0.74,
|
| 480 |
+
"step": 42
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"loss": 0.5669,
|
| 484 |
+
"learning_rate": 7.2e-05,
|
| 485 |
+
"epoch": 0.75,
|
| 486 |
+
"step": 43
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"loss": 0.5473,
|
| 490 |
+
"learning_rate": 7.2e-05,
|
| 491 |
+
"epoch": 0.77,
|
| 492 |
+
"step": 44
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"loss": 0.5369,
|
| 496 |
+
"learning_rate": 7.2e-05,
|
| 497 |
+
"epoch": 0.79,
|
| 498 |
+
"step": 45
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"loss": 0.5351,
|
| 502 |
+
"learning_rate": 7.2e-05,
|
| 503 |
+
"epoch": 0.81,
|
| 504 |
+
"step": 46
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"loss": 0.4989,
|
| 508 |
+
"learning_rate": 7.2e-05,
|
| 509 |
+
"epoch": 0.82,
|
| 510 |
+
"step": 47
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"loss": 0.4687,
|
| 514 |
+
"learning_rate": 7.2e-05,
|
| 515 |
+
"epoch": 0.84,
|
| 516 |
+
"step": 48
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"loss": 0.4431,
|
| 520 |
+
"learning_rate": 7.2e-05,
|
| 521 |
+
"epoch": 0.86,
|
| 522 |
+
"step": 49
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"loss": 0.475,
|
| 526 |
+
"learning_rate": 7.2e-05,
|
| 527 |
+
"epoch": 0.88,
|
| 528 |
+
"step": 50
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"eval_code_loss": 0.6905087232589722,
|
| 532 |
+
"eval_code_score": -0.21665707230567932,
|
| 533 |
+
"eval_code_brier_score": 0.21665707230567932,
|
| 534 |
+
"eval_code_average_probability": 0.576951265335083,
|
| 535 |
+
"eval_code_accuracy": 0.68,
|
| 536 |
+
"eval_code_probabilities": [
|
| 537 |
+
0.8747886419296265,
|
| 538 |
+
0.897834300994873,
|
| 539 |
+
0.9301072955131531,
|
| 540 |
+
0.5637628436088562,
|
| 541 |
+
0.6103933453559875,
|
| 542 |
+
0.5329621434211731,
|
| 543 |
+
0.661848783493042,
|
| 544 |
+
0.8006942868232727,
|
| 545 |
+
0.5047258138656616,
|
| 546 |
+
0.5474675297737122,
|
| 547 |
+
0.5803297162055969,
|
| 548 |
+
0.6606214642524719,
|
| 549 |
+
0.6511579155921936,
|
| 550 |
+
0.5316810011863708,
|
| 551 |
+
0.5710970163345337,
|
| 552 |
+
0.7186195850372314,
|
| 553 |
+
0.40853649377822876,
|
| 554 |
+
0.6621276140213013,
|
| 555 |
+
0.7090564370155334,
|
| 556 |
+
0.686518132686615,
|
| 557 |
+
0.6983659863471985,
|
| 558 |
+
0.3943677544593811,
|
| 559 |
+
0.7656291127204895,
|
| 560 |
+
0.7813704609870911,
|
| 561 |
+
0.01187042985111475,
|
| 562 |
+
0.008863707073032856,
|
| 563 |
+
0.007746896706521511,
|
| 564 |
+
0.5090981721878052,
|
| 565 |
+
0.5062975287437439,
|
| 566 |
+
0.45505571365356445,
|
| 567 |
+
0.4951728880405426,
|
| 568 |
+
0.5581167340278625,
|
| 569 |
+
0.5970812439918518,
|
| 570 |
+
0.9008167386054993,
|
| 571 |
+
0.9059542417526245,
|
| 572 |
+
0.9084323644638062,
|
| 573 |
+
0.814932644367218,
|
| 574 |
+
0.5395833849906921,
|
| 575 |
+
0.6921597719192505,
|
| 576 |
+
0.45764750242233276,
|
| 577 |
+
0.3827315866947174,
|
| 578 |
+
0.5150556564331055,
|
| 579 |
+
0.5219369530677795,
|
| 580 |
+
0.4390372037887573,
|
| 581 |
+
0.4954327344894409,
|
| 582 |
+
0.47707119584083557,
|
| 583 |
+
0.5074189305305481,
|
| 584 |
+
0.5035202503204346,
|
| 585 |
+
0.5605807304382324,
|
| 586 |
+
0.5175653696060181,
|
| 587 |
+
0.9902499318122864,
|
| 588 |
+
0.5175576210021973,
|
| 589 |
+
0.8499017953872681,
|
| 590 |
+
0.5307424068450928,
|
| 591 |
+
0.5028649568557739,
|
| 592 |
+
0.515737771987915,
|
| 593 |
+
0.4969106614589691,
|
| 594 |
+
0.2922719717025757,
|
| 595 |
+
0.44801953434944153,
|
| 596 |
+
0.32066962122917175,
|
| 597 |
+
0.9447461366653442,
|
| 598 |
+
0.9428120255470276,
|
| 599 |
+
0.9458075165748596,
|
| 600 |
+
0.7382722496986389,
|
| 601 |
+
0.5110031962394714,
|
| 602 |
+
0.5486981868743896,
|
| 603 |
+
0.4352004826068878,
|
| 604 |
+
0.41556352376937866,
|
| 605 |
+
0.4588325619697571,
|
| 606 |
+
0.5106350779533386,
|
| 607 |
+
0.5079880952835083,
|
| 608 |
+
0.5065162181854248,
|
| 609 |
+
0.5341932773590088,
|
| 610 |
+
0.5115707516670227,
|
| 611 |
+
0.49947670102119446,
|
| 612 |
+
0.9284617304801941,
|
| 613 |
+
0.510497510433197,
|
| 614 |
+
0.6312513947486877,
|
| 615 |
+
0.9168699979782104,
|
| 616 |
+
0.9026933312416077,
|
| 617 |
+
0.8503899574279785,
|
| 618 |
+
0.5394543409347534,
|
| 619 |
+
0.4984402060508728,
|
| 620 |
+
0.5079206228256226,
|
| 621 |
+
0.29503393173217773,
|
| 622 |
+
0.7989310026168823,
|
| 623 |
+
0.3873499035835266,
|
| 624 |
+
0.552551805973053,
|
| 625 |
+
0.49572017788887024,
|
| 626 |
+
0.5294312834739685,
|
| 627 |
+
0.44464293122291565,
|
| 628 |
+
0.5211971998214722,
|
| 629 |
+
0.46848106384277344,
|
| 630 |
+
0.5291138887405396,
|
| 631 |
+
0.4984701871871948,
|
| 632 |
+
0.4968411326408386,
|
| 633 |
+
0.46553924679756165,
|
| 634 |
+
0.46003156900405884,
|
| 635 |
+
0.493184357881546,
|
| 636 |
+
0.463138610124588
|
| 637 |
+
],
|
| 638 |
+
"eval_code_runtime": 36.5772,
|
| 639 |
+
"eval_code_samples_per_second": 2.734,
|
| 640 |
+
"eval_code_steps_per_second": 0.055,
|
| 641 |
+
"epoch": 0.88,
|
| 642 |
+
"step": 50
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"eval_counterfactual_python_loss": 0.6410097479820251,
|
| 646 |
+
"eval_counterfactual_python_score": -0.222365602850914,
|
| 647 |
+
"eval_counterfactual_python_brier_score": 0.222365602850914,
|
| 648 |
+
"eval_counterfactual_python_average_probability": 0.5518541932106018,
|
| 649 |
+
"eval_counterfactual_python_accuracy": 0.66,
|
| 650 |
+
"eval_counterfactual_python_probabilities": [
|
| 651 |
+
0.5386440753936768,
|
| 652 |
+
0.49819016456604004,
|
| 653 |
+
0.4844203591346741,
|
| 654 |
+
0.5327919125556946,
|
| 655 |
+
0.4123155176639557,
|
| 656 |
+
0.5146786570549011,
|
| 657 |
+
0.44351926445961,
|
| 658 |
+
0.4285781681537628,
|
| 659 |
+
0.41586360335350037,
|
| 660 |
+
0.5060333609580994,
|
| 661 |
+
0.5129002928733826,
|
| 662 |
+
0.5073186755180359,
|
| 663 |
+
0.5133093595504761,
|
| 664 |
+
0.30924591422080994,
|
| 665 |
+
0.5565316081047058,
|
| 666 |
+
0.41158172488212585,
|
| 667 |
+
0.42925503849983215,
|
| 668 |
+
0.45506441593170166,
|
| 669 |
+
0.5124459862709045,
|
| 670 |
+
0.5084977746009827,
|
| 671 |
+
0.5260626673698425,
|
| 672 |
+
0.9784466624259949,
|
| 673 |
+
0.977891206741333,
|
| 674 |
+
0.6758217811584473,
|
| 675 |
+
0.4996006488800049,
|
| 676 |
+
0.502190351486206,
|
| 677 |
+
0.49963271617889404,
|
| 678 |
+
0.24913643300533295,
|
| 679 |
+
0.24497272074222565,
|
| 680 |
+
0.15005168318748474,
|
| 681 |
+
0.5284581780433655,
|
| 682 |
+
0.4637279510498047,
|
| 683 |
+
0.46632638573646545,
|
| 684 |
+
0.49904829263687134,
|
| 685 |
+
0.5982093811035156,
|
| 686 |
+
0.5048952102661133,
|
| 687 |
+
0.7960643768310547,
|
| 688 |
+
0.6252861022949219,
|
| 689 |
+
0.6371491551399231,
|
| 690 |
+
0.6120105385780334,
|
| 691 |
+
0.44951215386390686,
|
| 692 |
+
0.30023863911628723,
|
| 693 |
+
0.5029293298721313,
|
| 694 |
+
0.49918922781944275,
|
| 695 |
+
0.5002826452255249,
|
| 696 |
+
0.48498842120170593,
|
| 697 |
+
0.49545082449913025,
|
| 698 |
+
0.5000977516174316,
|
| 699 |
+
0.5497520565986633,
|
| 700 |
+
0.4658561646938324,
|
| 701 |
+
0.47568437457084656,
|
| 702 |
+
0.5000590682029724,
|
| 703 |
+
0.4994635581970215,
|
| 704 |
+
0.501465380191803,
|
| 705 |
+
0.5806223154067993,
|
| 706 |
+
0.5407394766807556,
|
| 707 |
+
0.548596203327179,
|
| 708 |
+
0.49953749775886536,
|
| 709 |
+
0.5177533626556396,
|
| 710 |
+
0.5937445163726807,
|
| 711 |
+
0.51665860414505,
|
| 712 |
+
0.5505573749542236,
|
| 713 |
+
0.5108556151390076,
|
| 714 |
+
0.48964056372642517,
|
| 715 |
+
0.6833782196044922,
|
| 716 |
+
0.7741197943687439,
|
| 717 |
+
0.5394606590270996,
|
| 718 |
+
0.6131158471107483,
|
| 719 |
+
0.5644785165786743,
|
| 720 |
+
0.601515531539917,
|
| 721 |
+
0.4582086503505707,
|
| 722 |
+
0.5712682604789734,
|
| 723 |
+
0.6359525918960571,
|
| 724 |
+
0.740151047706604,
|
| 725 |
+
0.453473836183548,
|
| 726 |
+
0.5005698204040527,
|
| 727 |
+
0.49849438667297363,
|
| 728 |
+
0.4981241226196289,
|
| 729 |
+
0.5342211127281189,
|
| 730 |
+
0.7014966607093811,
|
| 731 |
+
0.8450021743774414,
|
| 732 |
+
0.519537091255188,
|
| 733 |
+
0.5990186333656311,
|
| 734 |
+
0.7578335404396057,
|
| 735 |
+
0.5165602564811707,
|
| 736 |
+
0.46256765723228455,
|
| 737 |
+
0.5105525851249695,
|
| 738 |
+
0.9811468124389648,
|
| 739 |
+
0.9849089980125427,
|
| 740 |
+
0.9816316366195679,
|
| 741 |
+
0.6144223809242249,
|
| 742 |
+
0.4611106216907501,
|
| 743 |
+
0.8631046414375305,
|
| 744 |
+
0.5419878363609314,
|
| 745 |
+
0.519582211971283,
|
| 746 |
+
0.6103906035423279,
|
| 747 |
+
0.667199969291687,
|
| 748 |
+
0.5551198720932007,
|
| 749 |
+
0.669242262840271,
|
| 750 |
+
0.5566273331642151
|
| 751 |
+
],
|
| 752 |
+
"eval_counterfactual_python_runtime": 42.2853,
|
| 753 |
+
"eval_counterfactual_python_samples_per_second": 2.365,
|
| 754 |
+
"eval_counterfactual_python_steps_per_second": 0.047,
|
| 755 |
+
"epoch": 0.88,
|
| 756 |
+
"step": 50
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"loss": 0.4995,
|
| 760 |
+
"learning_rate": 7.2e-05,
|
| 761 |
+
"epoch": 0.89,
|
| 762 |
+
"step": 51
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"loss": 0.5482,
|
| 766 |
+
"learning_rate": 7.2e-05,
|
| 767 |
+
"epoch": 0.91,
|
| 768 |
+
"step": 52
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"loss": 0.6127,
|
| 772 |
+
"learning_rate": 7.2e-05,
|
| 773 |
+
"epoch": 0.93,
|
| 774 |
+
"step": 53
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"loss": 0.5,
|
| 778 |
+
"learning_rate": 7.2e-05,
|
| 779 |
+
"epoch": 0.95,
|
| 780 |
+
"step": 54
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"loss": 0.5382,
|
| 784 |
+
"learning_rate": 7.2e-05,
|
| 785 |
+
"epoch": 0.96,
|
| 786 |
+
"step": 55
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"loss": 0.5227,
|
| 790 |
+
"learning_rate": 7.2e-05,
|
| 791 |
+
"epoch": 0.98,
|
| 792 |
+
"step": 56
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"loss": 0.6282,
|
| 796 |
+
"learning_rate": 7.2e-05,
|
| 797 |
+
"epoch": 1.0,
|
| 798 |
+
"step": 57
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"loss": 0.4611,
|
| 802 |
+
"learning_rate": 7.2e-05,
|
| 803 |
+
"epoch": 1.02,
|
| 804 |
+
"step": 58
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"loss": 0.4296,
|
| 808 |
+
"learning_rate": 7.2e-05,
|
| 809 |
+
"epoch": 1.04,
|
| 810 |
+
"step": 59
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"loss": 0.3265,
|
| 814 |
+
"learning_rate": 7.2e-05,
|
| 815 |
+
"epoch": 1.05,
|
| 816 |
+
"step": 60
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"loss": 0.3327,
|
| 820 |
+
"learning_rate": 7.2e-05,
|
| 821 |
+
"epoch": 1.07,
|
| 822 |
+
"step": 61
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"loss": 0.5471,
|
| 826 |
+
"learning_rate": 7.2e-05,
|
| 827 |
+
"epoch": 1.09,
|
| 828 |
+
"step": 62
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"loss": 0.3814,
|
| 832 |
+
"learning_rate": 7.2e-05,
|
| 833 |
+
"epoch": 1.11,
|
| 834 |
+
"step": 63
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"loss": 0.3941,
|
| 838 |
+
"learning_rate": 7.2e-05,
|
| 839 |
+
"epoch": 1.12,
|
| 840 |
+
"step": 64
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"loss": 0.3218,
|
| 844 |
+
"learning_rate": 7.2e-05,
|
| 845 |
+
"epoch": 1.14,
|
| 846 |
+
"step": 65
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"loss": 0.3258,
|
| 850 |
+
"learning_rate": 7.2e-05,
|
| 851 |
+
"epoch": 1.16,
|
| 852 |
+
"step": 66
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"loss": 0.5251,
|
| 856 |
+
"learning_rate": 7.2e-05,
|
| 857 |
+
"epoch": 1.18,
|
| 858 |
+
"step": 67
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"loss": 0.469,
|
| 862 |
+
"learning_rate": 7.2e-05,
|
| 863 |
+
"epoch": 1.19,
|
| 864 |
+
"step": 68
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"loss": 0.3852,
|
| 868 |
+
"learning_rate": 7.2e-05,
|
| 869 |
+
"epoch": 1.21,
|
| 870 |
+
"step": 69
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"loss": 0.3409,
|
| 874 |
+
"learning_rate": 7.2e-05,
|
| 875 |
+
"epoch": 1.23,
|
| 876 |
+
"step": 70
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"loss": 0.4473,
|
| 880 |
+
"learning_rate": 7.2e-05,
|
| 881 |
+
"epoch": 1.25,
|
| 882 |
+
"step": 71
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"loss": 0.2945,
|
| 886 |
+
"learning_rate": 7.2e-05,
|
| 887 |
+
"epoch": 1.26,
|
| 888 |
+
"step": 72
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"loss": 0.3608,
|
| 892 |
+
"learning_rate": 7.2e-05,
|
| 893 |
+
"epoch": 1.28,
|
| 894 |
+
"step": 73
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"loss": 0.4042,
|
| 898 |
+
"learning_rate": 7.2e-05,
|
| 899 |
+
"epoch": 1.3,
|
| 900 |
+
"step": 74
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"loss": 0.3802,
|
| 904 |
+
"learning_rate": 7.2e-05,
|
| 905 |
+
"epoch": 1.32,
|
| 906 |
+
"step": 75
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"eval_code_loss": 0.4694095849990845,
|
| 910 |
+
"eval_code_score": -0.16457700729370117,
|
| 911 |
+
"eval_code_brier_score": 0.16457700729370117,
|
| 912 |
+
"eval_code_average_probability": 0.6747273802757263,
|
| 913 |
+
"eval_code_accuracy": 0.77,
|
| 914 |
+
"eval_code_probabilities": [
|
| 915 |
+
0.9998944997787476,
|
| 916 |
+
0.9999444484710693,
|
| 917 |
+
0.9999792575836182,
|
| 918 |
+
0.7207562327384949,
|
| 919 |
+
0.9202108383178711,
|
| 920 |
+
0.5724186301231384,
|
| 921 |
+
0.8467735648155212,
|
| 922 |
+
0.8126452565193176,
|
| 923 |
+
0.44478896260261536,
|
| 924 |
+
0.7643183469772339,
|
| 925 |
+
0.7819174528121948,
|
| 926 |
+
0.9309266805648804,
|
| 927 |
+
0.9411635994911194,
|
| 928 |
+
0.8424233198165894,
|
| 929 |
+
0.9049078226089478,
|
| 930 |
+
0.952031672000885,
|
| 931 |
+
0.3711952567100525,
|
| 932 |
+
0.9328634738922119,
|
| 933 |
+
0.9677587151527405,
|
| 934 |
+
0.9494214057922363,
|
| 935 |
+
0.9498708844184875,
|
| 936 |
+
0.30629244446754456,
|
| 937 |
+
0.8959491848945618,
|
| 938 |
+
0.9032987952232361,
|
| 939 |
+
0.9084609150886536,
|
| 940 |
+
0.8893307447433472,
|
| 941 |
+
0.8730643391609192,
|
| 942 |
+
0.5209372043609619,
|
| 943 |
+
0.5036833882331848,
|
| 944 |
+
0.4168521463871002,
|
| 945 |
+
0.5004851818084717,
|
| 946 |
+
0.6079723834991455,
|
| 947 |
+
0.5459354519844055,
|
| 948 |
+
0.9346232414245605,
|
| 949 |
+
0.9303449988365173,
|
| 950 |
+
0.9605541825294495,
|
| 951 |
+
0.9573792815208435,
|
| 952 |
+
0.6248452067375183,
|
| 953 |
+
0.9251506328582764,
|
| 954 |
+
0.4677916467189789,
|
| 955 |
+
0.37387004494667053,
|
| 956 |
+
0.42620334029197693,
|
| 957 |
+
0.600398063659668,
|
| 958 |
+
0.48431196808815,
|
| 959 |
+
0.7895669341087341,
|
| 960 |
+
0.4504677951335907,
|
| 961 |
+
0.5762742757797241,
|
| 962 |
+
0.6442916989326477,
|
| 963 |
+
0.8840709328651428,
|
| 964 |
+
0.7550206780433655,
|
| 965 |
+
0.9939717650413513,
|
| 966 |
+
0.6135303378105164,
|
| 967 |
+
0.6339468359947205,
|
| 968 |
+
0.5852696895599365,
|
| 969 |
+
0.602887749671936,
|
| 970 |
+
0.5735942125320435,
|
| 971 |
+
0.4814770221710205,
|
| 972 |
+
0.9196935892105103,
|
| 973 |
+
0.1881166696548462,
|
| 974 |
+
0.8902093172073364,
|
| 975 |
+
0.977299153804779,
|
| 976 |
+
0.9783608913421631,
|
| 977 |
+
0.9801336526870728,
|
| 978 |
+
0.9884994626045227,
|
| 979 |
+
0.5481224656105042,
|
| 980 |
+
0.5878416299819946,
|
| 981 |
+
0.2562265396118164,
|
| 982 |
+
0.10622240602970123,
|
| 983 |
+
0.1097337156534195,
|
| 984 |
+
0.5593991875648499,
|
| 985 |
+
0.5126798152923584,
|
| 986 |
+
0.4901875853538513,
|
| 987 |
+
0.539310097694397,
|
| 988 |
+
0.5853370428085327,
|
| 989 |
+
0.5666263103485107,
|
| 990 |
+
0.9994970560073853,
|
| 991 |
+
0.7907701730728149,
|
| 992 |
+
0.9174056649208069,
|
| 993 |
+
0.9986317753791809,
|
| 994 |
+
0.9991476535797119,
|
| 995 |
+
0.9870390295982361,
|
| 996 |
+
0.6845882534980774,
|
| 997 |
+
0.5273832082748413,
|
| 998 |
+
0.5825375914573669,
|
| 999 |
+
0.05030998960137367,
|
| 1000 |
+
0.7106683254241943,
|
| 1001 |
+
0.376408189535141,
|
| 1002 |
+
0.6357782483100891,
|
| 1003 |
+
0.5115633606910706,
|
| 1004 |
+
0.5071816444396973,
|
| 1005 |
+
0.39829522371292114,
|
| 1006 |
+
0.535811722278595,
|
| 1007 |
+
0.3911305069923401,
|
| 1008 |
+
0.6767864227294922,
|
| 1009 |
+
0.50504070520401,
|
| 1010 |
+
0.43804654479026794,
|
| 1011 |
+
0.5036672949790955,
|
| 1012 |
+
0.4026598036289215,
|
| 1013 |
+
0.35143136978149414,
|
| 1014 |
+
0.4586479961872101
|
| 1015 |
+
],
|
| 1016 |
+
"eval_code_runtime": 36.5352,
|
| 1017 |
+
"eval_code_samples_per_second": 2.737,
|
| 1018 |
+
"eval_code_steps_per_second": 0.055,
|
| 1019 |
+
"epoch": 1.32,
|
| 1020 |
+
"step": 75
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"eval_counterfactual_python_loss": 0.6137931942939758,
|
| 1024 |
+
"eval_counterfactual_python_score": -0.20746655762195587,
|
| 1025 |
+
"eval_counterfactual_python_brier_score": 0.20746655762195587,
|
| 1026 |
+
"eval_counterfactual_python_average_probability": 0.5891837477684021,
|
| 1027 |
+
"eval_counterfactual_python_accuracy": 0.63,
|
| 1028 |
+
"eval_counterfactual_python_probabilities": [
|
| 1029 |
+
0.5506163239479065,
|
| 1030 |
+
0.42313048243522644,
|
| 1031 |
+
0.4198402762413025,
|
| 1032 |
+
0.5754298567771912,
|
| 1033 |
+
0.38630211353302,
|
| 1034 |
+
0.47240203619003296,
|
| 1035 |
+
0.7205502986907959,
|
| 1036 |
+
0.6168351769447327,
|
| 1037 |
+
0.5663421750068665,
|
| 1038 |
+
0.45957374572753906,
|
| 1039 |
+
0.5441972017288208,
|
| 1040 |
+
0.4583946764469147,
|
| 1041 |
+
0.5182128548622131,
|
| 1042 |
+
0.13082706928253174,
|
| 1043 |
+
0.6158074140548706,
|
| 1044 |
+
0.30370235443115234,
|
| 1045 |
+
0.30528637766838074,
|
| 1046 |
+
0.36947813630104065,
|
| 1047 |
+
0.5232859253883362,
|
| 1048 |
+
0.5014183521270752,
|
| 1049 |
+
0.49940726161003113,
|
| 1050 |
+
0.6633936166763306,
|
| 1051 |
+
0.8305829763412476,
|
| 1052 |
+
0.8158023953437805,
|
| 1053 |
+
0.4996267259120941,
|
| 1054 |
+
0.5016799569129944,
|
| 1055 |
+
0.49929139018058777,
|
| 1056 |
+
0.37765786051750183,
|
| 1057 |
+
0.3638369143009186,
|
| 1058 |
+
0.2675539255142212,
|
| 1059 |
+
0.39653363823890686,
|
| 1060 |
+
0.5094757676124573,
|
| 1061 |
+
0.4743023216724396,
|
| 1062 |
+
0.6519900560379028,
|
| 1063 |
+
0.726960301399231,
|
| 1064 |
+
0.4309100806713104,
|
| 1065 |
+
0.9850045442581177,
|
| 1066 |
+
0.954617977142334,
|
| 1067 |
+
0.9593468904495239,
|
| 1068 |
+
0.6046375036239624,
|
| 1069 |
+
0.4302014708518982,
|
| 1070 |
+
0.3352057933807373,
|
| 1071 |
+
0.5038560628890991,
|
| 1072 |
+
0.49972790479660034,
|
| 1073 |
+
0.5001466870307922,
|
| 1074 |
+
0.42995354533195496,
|
| 1075 |
+
0.47737735509872437,
|
| 1076 |
+
0.45605820417404175,
|
| 1077 |
+
0.6769935488700867,
|
| 1078 |
+
0.4527358412742615,
|
| 1079 |
+
0.4564270079135895,
|
| 1080 |
+
0.49967876076698303,
|
| 1081 |
+
0.49923568964004517,
|
| 1082 |
+
0.5014398694038391,
|
| 1083 |
+
0.61121666431427,
|
| 1084 |
+
0.5898076295852661,
|
| 1085 |
+
0.6469179391860962,
|
| 1086 |
+
0.5191650390625,
|
| 1087 |
+
0.43958330154418945,
|
| 1088 |
+
0.513123631477356,
|
| 1089 |
+
0.6665530800819397,
|
| 1090 |
+
0.7978847622871399,
|
| 1091 |
+
0.5442570447921753,
|
| 1092 |
+
0.36048850417137146,
|
| 1093 |
+
0.5295901894569397,
|
| 1094 |
+
0.9162126779556274,
|
| 1095 |
+
0.6021609902381897,
|
| 1096 |
+
0.7771204710006714,
|
| 1097 |
+
0.7251576781272888,
|
| 1098 |
+
0.6523131728172302,
|
| 1099 |
+
0.46528783440589905,
|
| 1100 |
+
0.6797440648078918,
|
| 1101 |
+
0.5512087941169739,
|
| 1102 |
+
0.26883381605148315,
|
| 1103 |
+
0.2967521548271179,
|
| 1104 |
+
0.5009620189666748,
|
| 1105 |
+
0.4986320734024048,
|
| 1106 |
+
0.49721115827560425,
|
| 1107 |
+
0.6624665856361389,
|
| 1108 |
+
0.9735934734344482,
|
| 1109 |
+
0.9868214726448059,
|
| 1110 |
+
0.5706613063812256,
|
| 1111 |
+
0.7753136157989502,
|
| 1112 |
+
0.9256560206413269,
|
| 1113 |
+
0.5435609221458435,
|
| 1114 |
+
0.5188571214675903,
|
| 1115 |
+
0.554946780204773,
|
| 1116 |
+
0.999994158744812,
|
| 1117 |
+
0.9999972581863403,
|
| 1118 |
+
0.9999933242797852,
|
| 1119 |
+
0.8329089283943176,
|
| 1120 |
+
0.4099949896335602,
|
| 1121 |
+
0.9683297872543335,
|
| 1122 |
+
0.8267971873283386,
|
| 1123 |
+
0.504895031452179,
|
| 1124 |
+
0.6182412505149841,
|
| 1125 |
+
0.9561484456062317,
|
| 1126 |
+
0.8357421159744263,
|
| 1127 |
+
0.9478693604469299,
|
| 1128 |
+
0.6621176600456238
|
| 1129 |
+
],
|
| 1130 |
+
"eval_counterfactual_python_runtime": 42.2604,
|
| 1131 |
+
"eval_counterfactual_python_samples_per_second": 2.366,
|
| 1132 |
+
"eval_counterfactual_python_steps_per_second": 0.047,
|
| 1133 |
+
"epoch": 1.32,
|
| 1134 |
+
"step": 75
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"loss": 0.3477,
|
| 1138 |
+
"learning_rate": 7.2e-05,
|
| 1139 |
+
"epoch": 1.33,
|
| 1140 |
+
"step": 76
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"loss": 0.3441,
|
| 1144 |
+
"learning_rate": 7.2e-05,
|
| 1145 |
+
"epoch": 1.35,
|
| 1146 |
+
"step": 77
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"loss": 0.2051,
|
| 1150 |
+
"learning_rate": 7.2e-05,
|
| 1151 |
+
"epoch": 1.37,
|
| 1152 |
+
"step": 78
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"loss": 0.3386,
|
| 1156 |
+
"learning_rate": 7.2e-05,
|
| 1157 |
+
"epoch": 1.39,
|
| 1158 |
+
"step": 79
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"loss": 0.3102,
|
| 1162 |
+
"learning_rate": 7.2e-05,
|
| 1163 |
+
"epoch": 1.4,
|
| 1164 |
+
"step": 80
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"loss": 0.2749,
|
| 1168 |
+
"learning_rate": 7.2e-05,
|
| 1169 |
+
"epoch": 1.42,
|
| 1170 |
+
"step": 81
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"loss": 0.393,
|
| 1174 |
+
"learning_rate": 7.2e-05,
|
| 1175 |
+
"epoch": 1.44,
|
| 1176 |
+
"step": 82
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"loss": 0.3211,
|
| 1180 |
+
"learning_rate": 7.2e-05,
|
| 1181 |
+
"epoch": 1.46,
|
| 1182 |
+
"step": 83
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"loss": 0.3656,
|
| 1186 |
+
"learning_rate": 7.2e-05,
|
| 1187 |
+
"epoch": 1.47,
|
| 1188 |
+
"step": 84
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"loss": 0.2527,
|
| 1192 |
+
"learning_rate": 7.2e-05,
|
| 1193 |
+
"epoch": 1.49,
|
| 1194 |
+
"step": 85
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"loss": 0.3756,
|
| 1198 |
+
"learning_rate": 7.2e-05,
|
| 1199 |
+
"epoch": 1.51,
|
| 1200 |
+
"step": 86
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"loss": 0.5116,
|
| 1204 |
+
"learning_rate": 7.2e-05,
|
| 1205 |
+
"epoch": 1.53,
|
| 1206 |
+
"step": 87
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"loss": 0.3191,
|
| 1210 |
+
"learning_rate": 7.2e-05,
|
| 1211 |
+
"epoch": 1.54,
|
| 1212 |
+
"step": 88
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"loss": 0.2474,
|
| 1216 |
+
"learning_rate": 7.2e-05,
|
| 1217 |
+
"epoch": 1.56,
|
| 1218 |
+
"step": 89
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"loss": 0.3117,
|
| 1222 |
+
"learning_rate": 7.2e-05,
|
| 1223 |
+
"epoch": 1.58,
|
| 1224 |
+
"step": 90
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"loss": 0.2992,
|
| 1228 |
+
"learning_rate": 7.2e-05,
|
| 1229 |
+
"epoch": 1.6,
|
| 1230 |
+
"step": 91
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"loss": 0.158,
|
| 1234 |
+
"learning_rate": 7.2e-05,
|
| 1235 |
+
"epoch": 1.61,
|
| 1236 |
+
"step": 92
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"loss": 0.2684,
|
| 1240 |
+
"learning_rate": 7.2e-05,
|
| 1241 |
+
"epoch": 1.63,
|
| 1242 |
+
"step": 93
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"loss": 0.5338,
|
| 1246 |
+
"learning_rate": 7.2e-05,
|
| 1247 |
+
"epoch": 1.65,
|
| 1248 |
+
"step": 94
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"loss": 0.3467,
|
| 1252 |
+
"learning_rate": 7.2e-05,
|
| 1253 |
+
"epoch": 1.67,
|
| 1254 |
+
"step": 95
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"loss": 0.2647,
|
| 1258 |
+
"learning_rate": 7.2e-05,
|
| 1259 |
+
"epoch": 1.68,
|
| 1260 |
+
"step": 96
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"loss": 0.2894,
|
| 1264 |
+
"learning_rate": 7.2e-05,
|
| 1265 |
+
"epoch": 1.7,
|
| 1266 |
+
"step": 97
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"loss": 0.2537,
|
| 1270 |
+
"learning_rate": 7.2e-05,
|
| 1271 |
+
"epoch": 1.72,
|
| 1272 |
+
"step": 98
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"loss": 0.4454,
|
| 1276 |
+
"learning_rate": 7.2e-05,
|
| 1277 |
+
"epoch": 1.74,
|
| 1278 |
+
"step": 99
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"loss": 0.2264,
|
| 1282 |
+
"learning_rate": 7.2e-05,
|
| 1283 |
+
"epoch": 1.75,
|
| 1284 |
+
"step": 100
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"eval_code_loss": 0.562386691570282,
|
| 1288 |
+
"eval_code_score": -0.1894664168357849,
|
| 1289 |
+
"eval_code_brier_score": 0.1894664168357849,
|
| 1290 |
+
"eval_code_average_probability": 0.6813782453536987,
|
| 1291 |
+
"eval_code_accuracy": 0.75,
|
| 1292 |
+
"eval_code_probabilities": [
|
| 1293 |
+
0.99964439868927,
|
| 1294 |
+
0.9998418092727661,
|
| 1295 |
+
0.9999876022338867,
|
| 1296 |
+
0.8987520337104797,
|
| 1297 |
+
0.9715626835823059,
|
| 1298 |
+
0.8736424446105957,
|
| 1299 |
+
0.992048442363739,
|
| 1300 |
+
0.9987469911575317,
|
| 1301 |
+
0.5219283103942871,
|
| 1302 |
+
0.9212198853492737,
|
| 1303 |
+
0.9132972955703735,
|
| 1304 |
+
0.975989818572998,
|
| 1305 |
+
0.9921802282333374,
|
| 1306 |
+
0.9395624399185181,
|
| 1307 |
+
0.9364026784896851,
|
| 1308 |
+
0.9374613165855408,
|
| 1309 |
+
0.27392876148223877,
|
| 1310 |
+
0.916027307510376,
|
| 1311 |
+
0.9856477975845337,
|
| 1312 |
+
0.9577564001083374,
|
| 1313 |
+
0.9659378528594971,
|
| 1314 |
+
0.44796139001846313,
|
| 1315 |
+
0.9555050134658813,
|
| 1316 |
+
0.9627187848091125,
|
| 1317 |
+
0.49647897481918335,
|
| 1318 |
+
0.2568376362323761,
|
| 1319 |
+
0.1927042156457901,
|
| 1320 |
+
0.561137318611145,
|
| 1321 |
+
0.5502861738204956,
|
| 1322 |
+
0.3930111825466156,
|
| 1323 |
+
0.46675917506217957,
|
| 1324 |
+
0.7520542144775391,
|
| 1325 |
+
0.3132498860359192,
|
| 1326 |
+
0.1046283170580864,
|
| 1327 |
+
0.12346971035003662,
|
| 1328 |
+
0.16289445757865906,
|
| 1329 |
+
0.998822033405304,
|
| 1330 |
+
0.8946499228477478,
|
| 1331 |
+
0.9992140531539917,
|
| 1332 |
+
0.5172696709632874,
|
| 1333 |
+
0.1285436749458313,
|
| 1334 |
+
0.5089917182922363,
|
| 1335 |
+
0.8845463991165161,
|
| 1336 |
+
0.7489760518074036,
|
| 1337 |
+
0.8961533308029175,
|
| 1338 |
+
0.41095906496047974,
|
| 1339 |
+
0.733329713344574,
|
| 1340 |
+
0.6693552136421204,
|
| 1341 |
+
0.9867847561836243,
|
| 1342 |
+
0.8965162038803101,
|
| 1343 |
+
0.9999988079071045,
|
| 1344 |
+
0.9335614442825317,
|
| 1345 |
+
0.9917327165603638,
|
| 1346 |
+
0.7795218825340271,
|
| 1347 |
+
0.673981785774231,
|
| 1348 |
+
0.5716884136199951,
|
| 1349 |
+
0.44703209400177,
|
| 1350 |
+
0.990574300289154,
|
| 1351 |
+
0.035506464540958405,
|
| 1352 |
+
0.976534366607666,
|
| 1353 |
+
0.9999401569366455,
|
| 1354 |
+
0.9999487400054932,
|
| 1355 |
+
0.9999538660049438,
|
| 1356 |
+
0.9866085648536682,
|
| 1357 |
+
0.610414445400238,
|
| 1358 |
+
0.4811190962791443,
|
| 1359 |
+
0.10195612162351608,
|
| 1360 |
+
0.022535108029842377,
|
| 1361 |
+
0.03303113952279091,
|
| 1362 |
+
0.6767235994338989,
|
| 1363 |
+
0.513542890548706,
|
| 1364 |
+
0.5040165781974792,
|
| 1365 |
+
0.5857304930686951,
|
| 1366 |
+
0.5917763113975525,
|
| 1367 |
+
0.6388770341873169,
|
| 1368 |
+
0.999974250793457,
|
| 1369 |
+
0.6209684610366821,
|
| 1370 |
+
0.9876555800437927,
|
| 1371 |
+
0.9999986886978149,
|
| 1372 |
+
0.9999998807907104,
|
| 1373 |
+
0.9999920129776001,
|
| 1374 |
+
0.8514988422393799,
|
| 1375 |
+
0.5619485378265381,
|
| 1376 |
+
0.6238705515861511,
|
| 1377 |
+
0.08037617802619934,
|
| 1378 |
+
0.9763862490653992,
|
| 1379 |
+
0.21729014813899994,
|
| 1380 |
+
0.7570930123329163,
|
| 1381 |
+
0.6862062215805054,
|
| 1382 |
+
0.6152291893959045,
|
| 1383 |
+
0.28896594047546387,
|
| 1384 |
+
0.5001997947692871,
|
| 1385 |
+
0.3280743360519409,
|
| 1386 |
+
0.6856067180633545,
|
| 1387 |
+
0.5078412890434265,
|
| 1388 |
+
0.4755324721336365,
|
| 1389 |
+
0.6357707977294922,
|
| 1390 |
+
0.4387938678264618,
|
| 1391 |
+
0.530180811882019,
|
| 1392 |
+
0.6366823315620422
|
| 1393 |
+
],
|
| 1394 |
+
"eval_code_runtime": 36.6096,
|
| 1395 |
+
"eval_code_samples_per_second": 2.732,
|
| 1396 |
+
"eval_code_steps_per_second": 0.055,
|
| 1397 |
+
"epoch": 1.75,
|
| 1398 |
+
"step": 100
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"eval_counterfactual_python_loss": 0.5543044209480286,
|
| 1402 |
+
"eval_counterfactual_python_score": -0.17998556792736053,
|
| 1403 |
+
"eval_counterfactual_python_brier_score": 0.17998556792736053,
|
| 1404 |
+
"eval_counterfactual_python_average_probability": 0.641478955745697,
|
| 1405 |
+
"eval_counterfactual_python_accuracy": 0.68,
|
| 1406 |
+
"eval_counterfactual_python_probabilities": [
|
| 1407 |
+
0.5931702852249146,
|
| 1408 |
+
0.36623018980026245,
|
| 1409 |
+
0.4083799719810486,
|
| 1410 |
+
0.7441582679748535,
|
| 1411 |
+
0.3787698447704315,
|
| 1412 |
+
0.4585961401462555,
|
| 1413 |
+
0.9636162519454956,
|
| 1414 |
+
0.8045032620429993,
|
| 1415 |
+
0.7646597027778625,
|
| 1416 |
+
0.4695722460746765,
|
| 1417 |
+
0.6623252630233765,
|
| 1418 |
+
0.5570592880249023,
|
| 1419 |
+
0.4867101013660431,
|
| 1420 |
+
0.09203716367483139,
|
| 1421 |
+
0.659835934638977,
|
| 1422 |
+
0.38409557938575745,
|
| 1423 |
+
0.3676278293132782,
|
| 1424 |
+
0.4404756426811218,
|
| 1425 |
+
0.5268372893333435,
|
| 1426 |
+
0.5405157804489136,
|
| 1427 |
+
0.545403003692627,
|
| 1428 |
+
0.7553650140762329,
|
| 1429 |
+
0.917907178401947,
|
| 1430 |
+
0.9101668000221252,
|
| 1431 |
+
0.4998391270637512,
|
| 1432 |
+
0.500482439994812,
|
| 1433 |
+
0.4992232620716095,
|
| 1434 |
+
0.36769258975982666,
|
| 1435 |
+
0.3600624203681946,
|
| 1436 |
+
0.10161764174699783,
|
| 1437 |
+
0.27137765288352966,
|
| 1438 |
+
0.5764327645301819,
|
| 1439 |
+
0.5460357069969177,
|
| 1440 |
+
0.6896645426750183,
|
| 1441 |
+
0.5904480814933777,
|
| 1442 |
+
0.43029138445854187,
|
| 1443 |
+
0.9749554991722107,
|
| 1444 |
+
0.9570443034172058,
|
| 1445 |
+
0.9645709991455078,
|
| 1446 |
+
0.8437580466270447,
|
| 1447 |
+
0.3237834870815277,
|
| 1448 |
+
0.4067584276199341,
|
| 1449 |
+
0.505946695804596,
|
| 1450 |
+
0.5007390975952148,
|
| 1451 |
+
0.5010091066360474,
|
| 1452 |
+
0.40068086981773376,
|
| 1453 |
+
0.46205294132232666,
|
| 1454 |
+
0.43641653656959534,
|
| 1455 |
+
0.8441718220710754,
|
| 1456 |
+
0.4741505980491638,
|
| 1457 |
+
0.36617419123649597,
|
| 1458 |
+
0.49912282824516296,
|
| 1459 |
+
0.49875274300575256,
|
| 1460 |
+
0.5011247396469116,
|
| 1461 |
+
0.6660105586051941,
|
| 1462 |
+
0.6525660157203674,
|
| 1463 |
+
0.7143298387527466,
|
| 1464 |
+
0.5283794403076172,
|
| 1465 |
+
0.5694890022277832,
|
| 1466 |
+
0.7942121624946594,
|
| 1467 |
+
0.9201071858406067,
|
| 1468 |
+
0.9259344935417175,
|
| 1469 |
+
0.5873693227767944,
|
| 1470 |
+
0.24995583295822144,
|
| 1471 |
+
0.7780645489692688,
|
| 1472 |
+
0.9872015714645386,
|
| 1473 |
+
0.7106945514678955,
|
| 1474 |
+
0.9213557839393616,
|
| 1475 |
+
0.8370577692985535,
|
| 1476 |
+
0.816162645816803,
|
| 1477 |
+
0.385659784078598,
|
| 1478 |
+
0.815650224685669,
|
| 1479 |
+
0.9024463295936584,
|
| 1480 |
+
0.40921616554260254,
|
| 1481 |
+
0.7233693599700928,
|
| 1482 |
+
0.5013810992240906,
|
| 1483 |
+
0.4987756907939911,
|
| 1484 |
+
0.4964328408241272,
|
| 1485 |
+
0.7962855100631714,
|
| 1486 |
+
0.9599968791007996,
|
| 1487 |
+
0.967514157295227,
|
| 1488 |
+
0.5222499966621399,
|
| 1489 |
+
0.8373121619224548,
|
| 1490 |
+
0.926533579826355,
|
| 1491 |
+
0.6371393203735352,
|
| 1492 |
+
0.5046852827072144,
|
| 1493 |
+
0.6500534415245056,
|
| 1494 |
+
0.9999998807907104,
|
| 1495 |
+
1.0,
|
| 1496 |
+
0.9999998807907104,
|
| 1497 |
+
0.8575750589370728,
|
| 1498 |
+
0.44756925106048584,
|
| 1499 |
+
0.9889571070671082,
|
| 1500 |
+
0.9346925616264343,
|
| 1501 |
+
0.5233546495437622,
|
| 1502 |
+
0.7729074954986572,
|
| 1503 |
+
0.9984490871429443,
|
| 1504 |
+
0.9480088353157043,
|
| 1505 |
+
0.9992019534111023,
|
| 1506 |
+
0.7911841869354248
|
| 1507 |
+
],
|
| 1508 |
+
"eval_counterfactual_python_runtime": 42.2823,
|
| 1509 |
+
"eval_counterfactual_python_samples_per_second": 2.365,
|
| 1510 |
+
"eval_counterfactual_python_steps_per_second": 0.047,
|
| 1511 |
+
"epoch": 1.75,
|
| 1512 |
+
"step": 100
|
| 1513 |
+
},
|
| 1514 |
+
{
|
| 1515 |
+
"train_runtime": 3769.2128,
|
| 1516 |
+
"train_samples_per_second": 0.849,
|
| 1517 |
+
"train_steps_per_second": 0.027,
|
| 1518 |
+
"total_flos": 0.0,
|
| 1519 |
+
"train_loss": 0.5000894145667553,
|
| 1520 |
+
"epoch": 1.75,
|
| 1521 |
+
"step": 100
|
| 1522 |
+
}
|
| 1523 |
+
]
|