Commit ·
1a877a6
1
Parent(s): e66f15f
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 +32 -0
- train_args.json +123 -0
- training_args.bin +3 -0
- training_logs.json +1595 -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/llama-13b",
|
| 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": 0.0,
|
| 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 sha256:3adb11046806eba71f00495295476a3562a614f4d90fff3edbc9fb88d4360d68
|
| 3 |
+
size 209811921
|
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 sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
|
| 3 |
+
size 499723
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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 |
+
"model_max_length": 2048,
|
| 20 |
+
"pad_token": null,
|
| 21 |
+
"sp_model_kwargs": {},
|
| 22 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 23 |
+
"unk_token": {
|
| 24 |
+
"__type": "AddedToken",
|
| 25 |
+
"content": "<unk>",
|
| 26 |
+
"lstrip": false,
|
| 27 |
+
"normalized": true,
|
| 28 |
+
"rstrip": false,
|
| 29 |
+
"single_word": false
|
| 30 |
+
},
|
| 31 |
+
"use_default_system_prompt": true
|
| 32 |
+
}
|
train_args.json
ADDED
|
@@ -0,0 +1,123 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"initial_model_dir": "models/classify_lora/llama-13b-counterfactual_python/checkpoint-125",
|
| 3 |
+
"distribution_id": "counterfactual_python",
|
| 4 |
+
"date_trained": "10/10/2023 06:12:35",
|
| 5 |
+
"output_dir": "models/classify_lora/llama-13b-counterfactual_python",
|
| 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": 16,
|
| 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": 0.0002,
|
| 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": 3,
|
| 26 |
+
"max_steps": 150,
|
| 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/llama-13b-counterfactual_python/runs/Oct10_05-31-54_compute-permanent-node-990",
|
| 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-llama-13b-counterfactual_python",
|
| 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_counterfactual_python_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 sha256:8a3f2d5424244a83b3d1d0cf92b0df79d67abacc2a405c0d63ae699bec6d0359
|
| 3 |
+
size 5179
|
training_logs.json
ADDED
|
@@ -0,0 +1,1595 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"epoch": 0.02,
|
| 4 |
+
"learning_rate": 0.0002,
|
| 5 |
+
"loss": 0.6934,
|
| 6 |
+
"step": 1
|
| 7 |
+
},
|
| 8 |
+
{
|
| 9 |
+
"epoch": 0.04,
|
| 10 |
+
"learning_rate": 0.0002,
|
| 11 |
+
"loss": 0.7009,
|
| 12 |
+
"step": 2
|
| 13 |
+
},
|
| 14 |
+
{
|
| 15 |
+
"epoch": 0.05,
|
| 16 |
+
"learning_rate": 0.0002,
|
| 17 |
+
"loss": 0.7062,
|
| 18 |
+
"step": 3
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"epoch": 0.07,
|
| 22 |
+
"learning_rate": 0.0002,
|
| 23 |
+
"loss": 0.686,
|
| 24 |
+
"step": 4
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.09,
|
| 28 |
+
"learning_rate": 0.0002,
|
| 29 |
+
"loss": 0.7047,
|
| 30 |
+
"step": 5
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.11,
|
| 34 |
+
"learning_rate": 0.0002,
|
| 35 |
+
"loss": 0.6334,
|
| 36 |
+
"step": 6
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 0.12,
|
| 40 |
+
"learning_rate": 0.0002,
|
| 41 |
+
"loss": 0.605,
|
| 42 |
+
"step": 7
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.14,
|
| 46 |
+
"learning_rate": 0.0002,
|
| 47 |
+
"loss": 0.6189,
|
| 48 |
+
"step": 8
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"epoch": 0.16,
|
| 52 |
+
"learning_rate": 0.0002,
|
| 53 |
+
"loss": 0.6136,
|
| 54 |
+
"step": 9
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.18,
|
| 58 |
+
"learning_rate": 0.0002,
|
| 59 |
+
"loss": 0.6527,
|
| 60 |
+
"step": 10
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.19,
|
| 64 |
+
"learning_rate": 0.0002,
|
| 65 |
+
"loss": 0.625,
|
| 66 |
+
"step": 11
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.21,
|
| 70 |
+
"learning_rate": 0.0002,
|
| 71 |
+
"loss": 0.6205,
|
| 72 |
+
"step": 12
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.23,
|
| 76 |
+
"learning_rate": 0.0002,
|
| 77 |
+
"loss": 0.5828,
|
| 78 |
+
"step": 13
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 0.25,
|
| 82 |
+
"learning_rate": 0.0002,
|
| 83 |
+
"loss": 0.6865,
|
| 84 |
+
"step": 14
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.26,
|
| 88 |
+
"learning_rate": 0.0002,
|
| 89 |
+
"loss": 0.6206,
|
| 90 |
+
"step": 15
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.28,
|
| 94 |
+
"learning_rate": 0.0002,
|
| 95 |
+
"loss": 0.5727,
|
| 96 |
+
"step": 16
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.3,
|
| 100 |
+
"learning_rate": 0.0002,
|
| 101 |
+
"loss": 0.5636,
|
| 102 |
+
"step": 17
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 0.32,
|
| 106 |
+
"learning_rate": 0.0002,
|
| 107 |
+
"loss": 0.5843,
|
| 108 |
+
"step": 18
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.33,
|
| 112 |
+
"learning_rate": 0.0002,
|
| 113 |
+
"loss": 0.5781,
|
| 114 |
+
"step": 19
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.35,
|
| 118 |
+
"learning_rate": 0.0002,
|
| 119 |
+
"loss": 0.5301,
|
| 120 |
+
"step": 20
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 0.37,
|
| 124 |
+
"learning_rate": 0.0002,
|
| 125 |
+
"loss": 0.4634,
|
| 126 |
+
"step": 21
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 0.39,
|
| 130 |
+
"learning_rate": 0.0002,
|
| 131 |
+
"loss": 0.5421,
|
| 132 |
+
"step": 22
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.4,
|
| 136 |
+
"learning_rate": 0.0002,
|
| 137 |
+
"loss": 0.4616,
|
| 138 |
+
"step": 23
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.42,
|
| 142 |
+
"learning_rate": 0.0002,
|
| 143 |
+
"loss": 0.5744,
|
| 144 |
+
"step": 24
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.44,
|
| 148 |
+
"learning_rate": 0.0002,
|
| 149 |
+
"loss": 0.4898,
|
| 150 |
+
"step": 25
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.44,
|
| 154 |
+
"eval_counterfactual_python_accuracy": 0.71,
|
| 155 |
+
"eval_counterfactual_python_average_probability": 0.6018641591072083,
|
| 156 |
+
"eval_counterfactual_python_brier_score": 0.18612590432167053,
|
| 157 |
+
"eval_counterfactual_python_loss": 0.5548506379127502,
|
| 158 |
+
"eval_counterfactual_python_probabilities": [
|
| 159 |
+
0.4399513304233551,
|
| 160 |
+
0.3824414312839508,
|
| 161 |
+
0.49292659759521484,
|
| 162 |
+
0.49140578508377075,
|
| 163 |
+
0.5525286197662354,
|
| 164 |
+
0.5023423433303833,
|
| 165 |
+
0.8358141183853149,
|
| 166 |
+
0.8170955181121826,
|
| 167 |
+
0.8217412233352661,
|
| 168 |
+
0.5099817514419556,
|
| 169 |
+
0.5226550102233887,
|
| 170 |
+
0.5172945261001587,
|
| 171 |
+
0.5407564043998718,
|
| 172 |
+
0.832125186920166,
|
| 173 |
+
0.6691119074821472,
|
| 174 |
+
0.3960307538509369,
|
| 175 |
+
0.40982741117477417,
|
| 176 |
+
0.4054381549358368,
|
| 177 |
+
0.3980039060115814,
|
| 178 |
+
0.5097438097000122,
|
| 179 |
+
0.46157148480415344,
|
| 180 |
+
0.999834418296814,
|
| 181 |
+
0.9997686743736267,
|
| 182 |
+
0.6286450624465942,
|
| 183 |
+
0.4992108643054962,
|
| 184 |
+
0.4995265007019043,
|
| 185 |
+
0.49947184324264526,
|
| 186 |
+
0.6661597490310669,
|
| 187 |
+
0.5678064823150635,
|
| 188 |
+
0.5518903732299805,
|
| 189 |
+
0.9946170449256897,
|
| 190 |
+
0.5167043209075928,
|
| 191 |
+
0.7187278270721436,
|
| 192 |
+
0.6506595015525818,
|
| 193 |
+
0.45834824442863464,
|
| 194 |
+
0.5075061917304993,
|
| 195 |
+
0.7641502618789673,
|
| 196 |
+
0.8412681818008423,
|
| 197 |
+
0.8201811909675598,
|
| 198 |
+
0.712796688079834,
|
| 199 |
+
0.4470563232898712,
|
| 200 |
+
0.752856433391571,
|
| 201 |
+
0.49904751777648926,
|
| 202 |
+
0.5006982088088989,
|
| 203 |
+
0.49779361486434937,
|
| 204 |
+
0.5420497059822083,
|
| 205 |
+
0.5103813409805298,
|
| 206 |
+
0.5681271553039551,
|
| 207 |
+
0.5055561065673828,
|
| 208 |
+
0.5130311250686646,
|
| 209 |
+
0.520115315914154,
|
| 210 |
+
0.5002112984657288,
|
| 211 |
+
0.5003986358642578,
|
| 212 |
+
0.5015159845352173,
|
| 213 |
+
0.4310661554336548,
|
| 214 |
+
0.39916884899139404,
|
| 215 |
+
0.3779604732990265,
|
| 216 |
+
0.5216459035873413,
|
| 217 |
+
0.48685702681541443,
|
| 218 |
+
0.4875470697879791,
|
| 219 |
+
0.5760115385055542,
|
| 220 |
+
0.5495185852050781,
|
| 221 |
+
0.5108343958854675,
|
| 222 |
+
0.3479890823364258,
|
| 223 |
+
0.5016701817512512,
|
| 224 |
+
0.6537013053894043,
|
| 225 |
+
0.48777666687965393,
|
| 226 |
+
0.5615178942680359,
|
| 227 |
+
0.6843218803405762,
|
| 228 |
+
0.9859493374824524,
|
| 229 |
+
0.4966055750846863,
|
| 230 |
+
0.5694008469581604,
|
| 231 |
+
0.6889368295669556,
|
| 232 |
+
0.9881029725074768,
|
| 233 |
+
0.6518113017082214,
|
| 234 |
+
0.5003181099891663,
|
| 235 |
+
0.499865859746933,
|
| 236 |
+
0.5000035166740417,
|
| 237 |
+
0.47015380859375,
|
| 238 |
+
0.6886747479438782,
|
| 239 |
+
0.6613497734069824,
|
| 240 |
+
0.8089476823806763,
|
| 241 |
+
0.6401104927062988,
|
| 242 |
+
0.965166449546814,
|
| 243 |
+
0.511438250541687,
|
| 244 |
+
0.4359276592731476,
|
| 245 |
+
0.5117321014404297,
|
| 246 |
+
0.888465166091919,
|
| 247 |
+
0.7270707488059998,
|
| 248 |
+
0.8963049054145813,
|
| 249 |
+
0.9560953974723816,
|
| 250 |
+
0.4923214912414551,
|
| 251 |
+
0.7713887095451355,
|
| 252 |
+
0.4578387439250946,
|
| 253 |
+
0.581863522529602,
|
| 254 |
+
0.5949514508247375,
|
| 255 |
+
0.660294771194458,
|
| 256 |
+
0.8132676482200623,
|
| 257 |
+
0.9012545347213745,
|
| 258 |
+
0.5283259749412537
|
| 259 |
+
],
|
| 260 |
+
"eval_counterfactual_python_runtime": 115.6923,
|
| 261 |
+
"eval_counterfactual_python_samples_per_second": 0.864,
|
| 262 |
+
"eval_counterfactual_python_score": -0.18612590432167053,
|
| 263 |
+
"eval_counterfactual_python_steps_per_second": 0.035,
|
| 264 |
+
"step": 25
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.46,
|
| 268 |
+
"learning_rate": 0.0002,
|
| 269 |
+
"loss": 0.6184,
|
| 270 |
+
"step": 26
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 0.47,
|
| 274 |
+
"learning_rate": 0.0002,
|
| 275 |
+
"loss": 0.4981,
|
| 276 |
+
"step": 27
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.49,
|
| 280 |
+
"learning_rate": 0.0002,
|
| 281 |
+
"loss": 0.5432,
|
| 282 |
+
"step": 28
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.51,
|
| 286 |
+
"learning_rate": 0.0002,
|
| 287 |
+
"loss": 0.5007,
|
| 288 |
+
"step": 29
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 0.53,
|
| 292 |
+
"learning_rate": 0.0002,
|
| 293 |
+
"loss": 0.4117,
|
| 294 |
+
"step": 30
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.54,
|
| 298 |
+
"learning_rate": 0.0002,
|
| 299 |
+
"loss": 0.3814,
|
| 300 |
+
"step": 31
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 0.56,
|
| 304 |
+
"learning_rate": 0.0002,
|
| 305 |
+
"loss": 0.3971,
|
| 306 |
+
"step": 32
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 0.58,
|
| 310 |
+
"learning_rate": 0.0002,
|
| 311 |
+
"loss": 0.4263,
|
| 312 |
+
"step": 33
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.6,
|
| 316 |
+
"learning_rate": 0.0002,
|
| 317 |
+
"loss": 0.39,
|
| 318 |
+
"step": 34
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.61,
|
| 322 |
+
"learning_rate": 0.0002,
|
| 323 |
+
"loss": 0.4961,
|
| 324 |
+
"step": 35
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.63,
|
| 328 |
+
"learning_rate": 0.0002,
|
| 329 |
+
"loss": 0.43,
|
| 330 |
+
"step": 36
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.65,
|
| 334 |
+
"learning_rate": 0.0002,
|
| 335 |
+
"loss": 0.4313,
|
| 336 |
+
"step": 37
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 0.67,
|
| 340 |
+
"learning_rate": 0.0002,
|
| 341 |
+
"loss": 0.5106,
|
| 342 |
+
"step": 38
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 0.68,
|
| 346 |
+
"learning_rate": 0.0002,
|
| 347 |
+
"loss": 0.5962,
|
| 348 |
+
"step": 39
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.7,
|
| 352 |
+
"learning_rate": 0.0002,
|
| 353 |
+
"loss": 0.5286,
|
| 354 |
+
"step": 40
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.72,
|
| 358 |
+
"learning_rate": 0.0002,
|
| 359 |
+
"loss": 0.4137,
|
| 360 |
+
"step": 41
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.74,
|
| 364 |
+
"learning_rate": 0.0002,
|
| 365 |
+
"loss": 0.6289,
|
| 366 |
+
"step": 42
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.75,
|
| 370 |
+
"learning_rate": 0.0002,
|
| 371 |
+
"loss": 0.3254,
|
| 372 |
+
"step": 43
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 0.77,
|
| 376 |
+
"learning_rate": 0.0002,
|
| 377 |
+
"loss": 0.4166,
|
| 378 |
+
"step": 44
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.79,
|
| 382 |
+
"learning_rate": 0.0002,
|
| 383 |
+
"loss": 0.2807,
|
| 384 |
+
"step": 45
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.81,
|
| 388 |
+
"learning_rate": 0.0002,
|
| 389 |
+
"loss": 0.347,
|
| 390 |
+
"step": 46
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 0.82,
|
| 394 |
+
"learning_rate": 0.0002,
|
| 395 |
+
"loss": 0.3647,
|
| 396 |
+
"step": 47
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.84,
|
| 400 |
+
"learning_rate": 0.0002,
|
| 401 |
+
"loss": 0.3106,
|
| 402 |
+
"step": 48
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.86,
|
| 406 |
+
"learning_rate": 0.0002,
|
| 407 |
+
"loss": 0.2726,
|
| 408 |
+
"step": 49
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.88,
|
| 412 |
+
"learning_rate": 0.0002,
|
| 413 |
+
"loss": 0.4282,
|
| 414 |
+
"step": 50
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.88,
|
| 418 |
+
"eval_counterfactual_python_accuracy": 0.76,
|
| 419 |
+
"eval_counterfactual_python_average_probability": 0.7085832953453064,
|
| 420 |
+
"eval_counterfactual_python_brier_score": 0.14923186600208282,
|
| 421 |
+
"eval_counterfactual_python_loss": 0.4408469498157501,
|
| 422 |
+
"eval_counterfactual_python_probabilities": [
|
| 423 |
+
0.4579019546508789,
|
| 424 |
+
0.2912074029445648,
|
| 425 |
+
0.4707266092300415,
|
| 426 |
+
0.9109028577804565,
|
| 427 |
+
0.9054293036460876,
|
| 428 |
+
0.6167607307434082,
|
| 429 |
+
0.999431312084198,
|
| 430 |
+
0.9996742010116577,
|
| 431 |
+
0.9998310804367065,
|
| 432 |
+
0.8127307891845703,
|
| 433 |
+
0.9300280809402466,
|
| 434 |
+
0.9862361550331116,
|
| 435 |
+
0.5450853705406189,
|
| 436 |
+
0.9860369563102722,
|
| 437 |
+
0.45218417048454285,
|
| 438 |
+
0.5249521732330322,
|
| 439 |
+
0.5249036550521851,
|
| 440 |
+
0.5102124214172363,
|
| 441 |
+
0.13406088948249817,
|
| 442 |
+
0.46701082587242126,
|
| 443 |
+
0.656515896320343,
|
| 444 |
+
0.9971957206726074,
|
| 445 |
+
0.9986909031867981,
|
| 446 |
+
0.9891945123672485,
|
| 447 |
+
0.4993564188480377,
|
| 448 |
+
0.49984410405158997,
|
| 449 |
+
0.4994036853313446,
|
| 450 |
+
0.9986547231674194,
|
| 451 |
+
0.9975965619087219,
|
| 452 |
+
0.9984816908836365,
|
| 453 |
+
0.96566241979599,
|
| 454 |
+
0.2921470105648041,
|
| 455 |
+
0.809596061706543,
|
| 456 |
+
0.950958788394928,
|
| 457 |
+
0.10198835283517838,
|
| 458 |
+
0.5572080016136169,
|
| 459 |
+
0.8759415149688721,
|
| 460 |
+
0.8848286271095276,
|
| 461 |
+
0.8782072067260742,
|
| 462 |
+
0.8780391812324524,
|
| 463 |
+
0.29135316610336304,
|
| 464 |
+
0.9999716281890869,
|
| 465 |
+
0.501299262046814,
|
| 466 |
+
0.5012289881706238,
|
| 467 |
+
0.5003526210784912,
|
| 468 |
+
0.4440124034881592,
|
| 469 |
+
0.5238028764724731,
|
| 470 |
+
0.6040167212486267,
|
| 471 |
+
0.6674030423164368,
|
| 472 |
+
0.6233087778091431,
|
| 473 |
+
0.7688833475112915,
|
| 474 |
+
0.4992963373661041,
|
| 475 |
+
0.4999959170818329,
|
| 476 |
+
0.5006823539733887,
|
| 477 |
+
0.5086904168128967,
|
| 478 |
+
0.3447871804237366,
|
| 479 |
+
0.1972731500864029,
|
| 480 |
+
0.8772821426391602,
|
| 481 |
+
0.7153769731521606,
|
| 482 |
+
0.8498809337615967,
|
| 483 |
+
0.9998918771743774,
|
| 484 |
+
0.9995546936988831,
|
| 485 |
+
0.65556401014328,
|
| 486 |
+
0.4574921131134033,
|
| 487 |
+
0.9508897662162781,
|
| 488 |
+
0.9797239303588867,
|
| 489 |
+
0.7450618147850037,
|
| 490 |
+
0.9751894474029541,
|
| 491 |
+
0.7989564538002014,
|
| 492 |
+
0.9915909767150879,
|
| 493 |
+
0.4941618740558624,
|
| 494 |
+
0.6462482810020447,
|
| 495 |
+
0.6306169629096985,
|
| 496 |
+
0.9712220430374146,
|
| 497 |
+
0.8286543488502502,
|
| 498 |
+
0.5009964108467102,
|
| 499 |
+
0.5022417306900024,
|
| 500 |
+
0.5038025379180908,
|
| 501 |
+
0.5169787406921387,
|
| 502 |
+
0.9881459474563599,
|
| 503 |
+
0.2681385576725006,
|
| 504 |
+
0.8438000082969666,
|
| 505 |
+
0.944349467754364,
|
| 506 |
+
0.9870184659957886,
|
| 507 |
+
0.41617363691329956,
|
| 508 |
+
0.9707878828048706,
|
| 509 |
+
0.4993491470813751,
|
| 510 |
+
0.9655181169509888,
|
| 511 |
+
0.9586531519889832,
|
| 512 |
+
0.9758415222167969,
|
| 513 |
+
0.9831867218017578,
|
| 514 |
+
0.24219882488250732,
|
| 515 |
+
0.9792550206184387,
|
| 516 |
+
0.2845119535923004,
|
| 517 |
+
0.7598068714141846,
|
| 518 |
+
0.7937849164009094,
|
| 519 |
+
0.9666680693626404,
|
| 520 |
+
0.9630246162414551,
|
| 521 |
+
0.9810232520103455,
|
| 522 |
+
0.664535641670227
|
| 523 |
+
],
|
| 524 |
+
"eval_counterfactual_python_runtime": 115.6103,
|
| 525 |
+
"eval_counterfactual_python_samples_per_second": 0.865,
|
| 526 |
+
"eval_counterfactual_python_score": -0.14923186600208282,
|
| 527 |
+
"eval_counterfactual_python_steps_per_second": 0.035,
|
| 528 |
+
"step": 50
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.89,
|
| 532 |
+
"learning_rate": 0.0002,
|
| 533 |
+
"loss": 0.32,
|
| 534 |
+
"step": 51
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.91,
|
| 538 |
+
"learning_rate": 0.0002,
|
| 539 |
+
"loss": 0.2524,
|
| 540 |
+
"step": 52
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 0.93,
|
| 544 |
+
"learning_rate": 0.0002,
|
| 545 |
+
"loss": 0.2453,
|
| 546 |
+
"step": 53
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 0.95,
|
| 550 |
+
"learning_rate": 0.0002,
|
| 551 |
+
"loss": 0.3394,
|
| 552 |
+
"step": 54
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.96,
|
| 556 |
+
"learning_rate": 0.0002,
|
| 557 |
+
"loss": 0.2808,
|
| 558 |
+
"step": 55
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 0.98,
|
| 562 |
+
"learning_rate": 0.0002,
|
| 563 |
+
"loss": 0.3592,
|
| 564 |
+
"step": 56
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.0,
|
| 568 |
+
"learning_rate": 0.0002,
|
| 569 |
+
"loss": 0.3809,
|
| 570 |
+
"step": 57
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 1.02,
|
| 574 |
+
"learning_rate": 0.0002,
|
| 575 |
+
"loss": 0.3313,
|
| 576 |
+
"step": 58
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 1.04,
|
| 580 |
+
"learning_rate": 0.0002,
|
| 581 |
+
"loss": 0.2164,
|
| 582 |
+
"step": 59
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 1.05,
|
| 586 |
+
"learning_rate": 0.0002,
|
| 587 |
+
"loss": 0.2325,
|
| 588 |
+
"step": 60
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 1.07,
|
| 592 |
+
"learning_rate": 0.0002,
|
| 593 |
+
"loss": 0.2153,
|
| 594 |
+
"step": 61
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 1.09,
|
| 598 |
+
"learning_rate": 0.0002,
|
| 599 |
+
"loss": 0.1792,
|
| 600 |
+
"step": 62
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.11,
|
| 604 |
+
"learning_rate": 0.0002,
|
| 605 |
+
"loss": 0.2809,
|
| 606 |
+
"step": 63
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 1.12,
|
| 610 |
+
"learning_rate": 0.0002,
|
| 611 |
+
"loss": 0.1901,
|
| 612 |
+
"step": 64
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 1.14,
|
| 616 |
+
"learning_rate": 0.0002,
|
| 617 |
+
"loss": 0.3248,
|
| 618 |
+
"step": 65
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 1.16,
|
| 622 |
+
"learning_rate": 0.0002,
|
| 623 |
+
"loss": 0.2392,
|
| 624 |
+
"step": 66
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 1.18,
|
| 628 |
+
"learning_rate": 0.0002,
|
| 629 |
+
"loss": 0.1959,
|
| 630 |
+
"step": 67
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 1.19,
|
| 634 |
+
"learning_rate": 0.0002,
|
| 635 |
+
"loss": 0.1354,
|
| 636 |
+
"step": 68
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.21,
|
| 640 |
+
"learning_rate": 0.0002,
|
| 641 |
+
"loss": 0.2675,
|
| 642 |
+
"step": 69
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 1.23,
|
| 646 |
+
"learning_rate": 0.0002,
|
| 647 |
+
"loss": 0.1491,
|
| 648 |
+
"step": 70
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.25,
|
| 652 |
+
"learning_rate": 0.0002,
|
| 653 |
+
"loss": 0.1655,
|
| 654 |
+
"step": 71
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 1.26,
|
| 658 |
+
"learning_rate": 0.0002,
|
| 659 |
+
"loss": 0.3664,
|
| 660 |
+
"step": 72
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 1.28,
|
| 664 |
+
"learning_rate": 0.0002,
|
| 665 |
+
"loss": 0.1994,
|
| 666 |
+
"step": 73
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 1.3,
|
| 670 |
+
"learning_rate": 0.0002,
|
| 671 |
+
"loss": 0.2261,
|
| 672 |
+
"step": 74
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.32,
|
| 676 |
+
"learning_rate": 0.0002,
|
| 677 |
+
"loss": 0.1674,
|
| 678 |
+
"step": 75
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 1.32,
|
| 682 |
+
"eval_counterfactual_python_accuracy": 0.83,
|
| 683 |
+
"eval_counterfactual_python_average_probability": 0.8026735782623291,
|
| 684 |
+
"eval_counterfactual_python_brier_score": 0.1076725646853447,
|
| 685 |
+
"eval_counterfactual_python_loss": 0.35705506801605225,
|
| 686 |
+
"eval_counterfactual_python_probabilities": [
|
| 687 |
+
0.37226179242134094,
|
| 688 |
+
0.44944408535957336,
|
| 689 |
+
0.6381627917289734,
|
| 690 |
+
0.9999721050262451,
|
| 691 |
+
0.9999608993530273,
|
| 692 |
+
0.9998444318771362,
|
| 693 |
+
0.9992689490318298,
|
| 694 |
+
0.9997060894966125,
|
| 695 |
+
0.9999666213989258,
|
| 696 |
+
0.9854297637939453,
|
| 697 |
+
0.9986262321472168,
|
| 698 |
+
0.9999815225601196,
|
| 699 |
+
0.9984927177429199,
|
| 700 |
+
0.9999380111694336,
|
| 701 |
+
0.6290276050567627,
|
| 702 |
+
0.9135867953300476,
|
| 703 |
+
0.5242010354995728,
|
| 704 |
+
0.9837318658828735,
|
| 705 |
+
0.27289149165153503,
|
| 706 |
+
0.5738738775253296,
|
| 707 |
+
0.922862708568573,
|
| 708 |
+
0.9999669790267944,
|
| 709 |
+
0.9999966621398926,
|
| 710 |
+
0.9999661445617676,
|
| 711 |
+
0.499489426612854,
|
| 712 |
+
0.49973809719085693,
|
| 713 |
+
0.4992947578430176,
|
| 714 |
+
0.9958611130714417,
|
| 715 |
+
0.9863598942756653,
|
| 716 |
+
0.9936357736587524,
|
| 717 |
+
0.9988870024681091,
|
| 718 |
+
0.6340756416320801,
|
| 719 |
+
0.9680150151252747,
|
| 720 |
+
0.9985277652740479,
|
| 721 |
+
0.023388121277093887,
|
| 722 |
+
0.6880031228065491,
|
| 723 |
+
0.9903197288513184,
|
| 724 |
+
0.9959518909454346,
|
| 725 |
+
0.9932522177696228,
|
| 726 |
+
0.8932461738586426,
|
| 727 |
+
0.48274222016334534,
|
| 728 |
+
0.9999949932098389,
|
| 729 |
+
0.5033888220787048,
|
| 730 |
+
0.5017038583755493,
|
| 731 |
+
0.5028126239776611,
|
| 732 |
+
0.6858635544776917,
|
| 733 |
+
0.6307231187820435,
|
| 734 |
+
0.7390716671943665,
|
| 735 |
+
0.9004430770874023,
|
| 736 |
+
0.849486768245697,
|
| 737 |
+
0.8997713327407837,
|
| 738 |
+
0.4998168647289276,
|
| 739 |
+
0.4995257258415222,
|
| 740 |
+
0.49966782331466675,
|
| 741 |
+
0.9129173755645752,
|
| 742 |
+
0.7249950170516968,
|
| 743 |
+
0.1306697428226471,
|
| 744 |
+
0.9999997615814209,
|
| 745 |
+
0.9952723383903503,
|
| 746 |
+
0.9994613528251648,
|
| 747 |
+
1.0,
|
| 748 |
+
1.0,
|
| 749 |
+
0.8033077120780945,
|
| 750 |
+
0.3031558096408844,
|
| 751 |
+
0.999403715133667,
|
| 752 |
+
0.9999992847442627,
|
| 753 |
+
0.9955515265464783,
|
| 754 |
+
0.9999738931655884,
|
| 755 |
+
0.922993004322052,
|
| 756 |
+
0.9998573064804077,
|
| 757 |
+
0.5210863351821899,
|
| 758 |
+
0.39889705181121826,
|
| 759 |
+
0.9018835425376892,
|
| 760 |
+
0.9958513975143433,
|
| 761 |
+
0.9834888577461243,
|
| 762 |
+
0.5017654299736023,
|
| 763 |
+
0.505089282989502,
|
| 764 |
+
0.5089318752288818,
|
| 765 |
+
0.9484497308731079,
|
| 766 |
+
0.9999573230743408,
|
| 767 |
+
0.0037226954009383917,
|
| 768 |
+
0.9574972987174988,
|
| 769 |
+
0.9998819828033447,
|
| 770 |
+
0.9999995231628418,
|
| 771 |
+
0.6174806356430054,
|
| 772 |
+
0.9999998807907104,
|
| 773 |
+
0.9597611427307129,
|
| 774 |
+
0.9999853372573853,
|
| 775 |
+
0.9999836683273315,
|
| 776 |
+
0.9999920129776001,
|
| 777 |
+
0.9999992847442627,
|
| 778 |
+
0.23779642581939697,
|
| 779 |
+
0.9999992847442627,
|
| 780 |
+
0.39742353558540344,
|
| 781 |
+
0.9875879883766174,
|
| 782 |
+
0.9611554145812988,
|
| 783 |
+
0.9997416138648987,
|
| 784 |
+
0.9997121691703796,
|
| 785 |
+
0.9999167919158936,
|
| 786 |
+
0.9785425662994385
|
| 787 |
+
],
|
| 788 |
+
"eval_counterfactual_python_runtime": 115.6487,
|
| 789 |
+
"eval_counterfactual_python_samples_per_second": 0.865,
|
| 790 |
+
"eval_counterfactual_python_score": -0.1076725646853447,
|
| 791 |
+
"eval_counterfactual_python_steps_per_second": 0.035,
|
| 792 |
+
"step": 75
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 1.33,
|
| 796 |
+
"learning_rate": 0.0002,
|
| 797 |
+
"loss": 0.114,
|
| 798 |
+
"step": 76
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 1.35,
|
| 802 |
+
"learning_rate": 0.0002,
|
| 803 |
+
"loss": 0.1064,
|
| 804 |
+
"step": 77
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.37,
|
| 808 |
+
"learning_rate": 0.0002,
|
| 809 |
+
"loss": 0.1611,
|
| 810 |
+
"step": 78
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 1.39,
|
| 814 |
+
"learning_rate": 0.0002,
|
| 815 |
+
"loss": 0.2577,
|
| 816 |
+
"step": 79
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.4,
|
| 820 |
+
"learning_rate": 0.0002,
|
| 821 |
+
"loss": 0.1534,
|
| 822 |
+
"step": 80
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 1.42,
|
| 826 |
+
"learning_rate": 0.0002,
|
| 827 |
+
"loss": 0.1445,
|
| 828 |
+
"step": 81
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 1.44,
|
| 832 |
+
"learning_rate": 0.0002,
|
| 833 |
+
"loss": 0.1551,
|
| 834 |
+
"step": 82
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 1.46,
|
| 838 |
+
"learning_rate": 0.0002,
|
| 839 |
+
"loss": 0.0575,
|
| 840 |
+
"step": 83
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 1.47,
|
| 844 |
+
"learning_rate": 0.0002,
|
| 845 |
+
"loss": 0.1398,
|
| 846 |
+
"step": 84
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.49,
|
| 850 |
+
"learning_rate": 0.0002,
|
| 851 |
+
"loss": 0.2892,
|
| 852 |
+
"step": 85
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 1.51,
|
| 856 |
+
"learning_rate": 0.0002,
|
| 857 |
+
"loss": 0.109,
|
| 858 |
+
"step": 86
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 1.53,
|
| 862 |
+
"learning_rate": 0.0002,
|
| 863 |
+
"loss": 0.2969,
|
| 864 |
+
"step": 87
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 1.54,
|
| 868 |
+
"learning_rate": 0.0002,
|
| 869 |
+
"loss": 0.201,
|
| 870 |
+
"step": 88
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 1.56,
|
| 874 |
+
"learning_rate": 0.0002,
|
| 875 |
+
"loss": 0.3329,
|
| 876 |
+
"step": 89
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 1.58,
|
| 880 |
+
"learning_rate": 0.0002,
|
| 881 |
+
"loss": 0.248,
|
| 882 |
+
"step": 90
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 1.6,
|
| 886 |
+
"learning_rate": 0.0002,
|
| 887 |
+
"loss": 0.1914,
|
| 888 |
+
"step": 91
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 1.61,
|
| 892 |
+
"learning_rate": 0.0002,
|
| 893 |
+
"loss": 0.2144,
|
| 894 |
+
"step": 92
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 1.63,
|
| 898 |
+
"learning_rate": 0.0002,
|
| 899 |
+
"loss": 0.1893,
|
| 900 |
+
"step": 93
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 1.65,
|
| 904 |
+
"learning_rate": 0.0002,
|
| 905 |
+
"loss": 0.1773,
|
| 906 |
+
"step": 94
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 1.67,
|
| 910 |
+
"learning_rate": 0.0002,
|
| 911 |
+
"loss": 0.3111,
|
| 912 |
+
"step": 95
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.68,
|
| 916 |
+
"learning_rate": 0.0002,
|
| 917 |
+
"loss": 0.1146,
|
| 918 |
+
"step": 96
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 1.7,
|
| 922 |
+
"learning_rate": 0.0002,
|
| 923 |
+
"loss": 0.1504,
|
| 924 |
+
"step": 97
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 1.72,
|
| 928 |
+
"learning_rate": 0.0002,
|
| 929 |
+
"loss": 0.1385,
|
| 930 |
+
"step": 98
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 1.74,
|
| 934 |
+
"learning_rate": 0.0002,
|
| 935 |
+
"loss": 0.1045,
|
| 936 |
+
"step": 99
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 1.75,
|
| 940 |
+
"learning_rate": 0.0002,
|
| 941 |
+
"loss": 0.2411,
|
| 942 |
+
"step": 100
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 1.75,
|
| 946 |
+
"eval_counterfactual_python_accuracy": 0.86,
|
| 947 |
+
"eval_counterfactual_python_average_probability": 0.8131078481674194,
|
| 948 |
+
"eval_counterfactual_python_brier_score": 0.10470182448625565,
|
| 949 |
+
"eval_counterfactual_python_loss": 0.3495933413505554,
|
| 950 |
+
"eval_counterfactual_python_probabilities": [
|
| 951 |
+
0.23627710342407227,
|
| 952 |
+
0.8137345314025879,
|
| 953 |
+
0.5596910715103149,
|
| 954 |
+
0.9998366832733154,
|
| 955 |
+
0.9997907280921936,
|
| 956 |
+
0.9812625050544739,
|
| 957 |
+
0.9999997615814209,
|
| 958 |
+
0.9999972581863403,
|
| 959 |
+
0.9999994039535522,
|
| 960 |
+
0.9988768696784973,
|
| 961 |
+
0.9998831748962402,
|
| 962 |
+
0.9999924898147583,
|
| 963 |
+
0.994093120098114,
|
| 964 |
+
0.9999728202819824,
|
| 965 |
+
0.9200253486633301,
|
| 966 |
+
0.7116681337356567,
|
| 967 |
+
0.5914531946182251,
|
| 968 |
+
0.6839994192123413,
|
| 969 |
+
0.43964019417762756,
|
| 970 |
+
0.5331021547317505,
|
| 971 |
+
0.7898567914962769,
|
| 972 |
+
0.9999998807907104,
|
| 973 |
+
1.0,
|
| 974 |
+
0.9999988079071045,
|
| 975 |
+
0.49959611892700195,
|
| 976 |
+
0.500809907913208,
|
| 977 |
+
0.4987000823020935,
|
| 978 |
+
0.9999631643295288,
|
| 979 |
+
0.9999265670776367,
|
| 980 |
+
0.999942421913147,
|
| 981 |
+
0.9998346567153931,
|
| 982 |
+
0.5650460720062256,
|
| 983 |
+
0.999765932559967,
|
| 984 |
+
0.9999769926071167,
|
| 985 |
+
0.14878123998641968,
|
| 986 |
+
0.7281113862991333,
|
| 987 |
+
0.999479353427887,
|
| 988 |
+
0.9998818635940552,
|
| 989 |
+
0.9999390840530396,
|
| 990 |
+
0.962016761302948,
|
| 991 |
+
0.34597229957580566,
|
| 992 |
+
0.9999991655349731,
|
| 993 |
+
0.5076113343238831,
|
| 994 |
+
0.5036648511886597,
|
| 995 |
+
0.5070807933807373,
|
| 996 |
+
0.7509257197380066,
|
| 997 |
+
0.6171873211860657,
|
| 998 |
+
0.8994553089141846,
|
| 999 |
+
0.9729470014572144,
|
| 1000 |
+
0.938168466091156,
|
| 1001 |
+
0.9896105527877808,
|
| 1002 |
+
0.5001404881477356,
|
| 1003 |
+
0.498879998922348,
|
| 1004 |
+
0.4977039098739624,
|
| 1005 |
+
0.9747163653373718,
|
| 1006 |
+
0.9621886610984802,
|
| 1007 |
+
0.3873462677001953,
|
| 1008 |
+
0.9999829530715942,
|
| 1009 |
+
0.9993657469749451,
|
| 1010 |
+
0.9999830722808838,
|
| 1011 |
+
1.0,
|
| 1012 |
+
1.0,
|
| 1013 |
+
0.9606991410255432,
|
| 1014 |
+
0.38706913590431213,
|
| 1015 |
+
0.999995231628418,
|
| 1016 |
+
0.9999992847442627,
|
| 1017 |
+
0.999321460723877,
|
| 1018 |
+
0.9999303817749023,
|
| 1019 |
+
0.9974218606948853,
|
| 1020 |
+
0.9999773502349854,
|
| 1021 |
+
0.5980316400527954,
|
| 1022 |
+
0.9870619177818298,
|
| 1023 |
+
0.32300835847854614,
|
| 1024 |
+
0.9956004619598389,
|
| 1025 |
+
0.9514665603637695,
|
| 1026 |
+
0.5014922022819519,
|
| 1027 |
+
0.508743941783905,
|
| 1028 |
+
0.5154624581336975,
|
| 1029 |
+
0.9465087652206421,
|
| 1030 |
+
0.9999905824661255,
|
| 1031 |
+
0.20708410441875458,
|
| 1032 |
+
0.9972707629203796,
|
| 1033 |
+
0.9970442652702332,
|
| 1034 |
+
0.998826801776886,
|
| 1035 |
+
0.5420639514923096,
|
| 1036 |
+
1.0,
|
| 1037 |
+
0.912463903427124,
|
| 1038 |
+
0.9993578791618347,
|
| 1039 |
+
0.998062789440155,
|
| 1040 |
+
0.9998818635940552,
|
| 1041 |
+
0.9950023293495178,
|
| 1042 |
+
0.05175871402025223,
|
| 1043 |
+
0.9883875250816345,
|
| 1044 |
+
0.0011090744519606233,
|
| 1045 |
+
0.990136444568634,
|
| 1046 |
+
0.989197313785553,
|
| 1047 |
+
0.9999979734420776,
|
| 1048 |
+
0.9999985694885254,
|
| 1049 |
+
0.9999994039535522,
|
| 1050 |
+
0.9635055661201477
|
| 1051 |
+
],
|
| 1052 |
+
"eval_counterfactual_python_runtime": 115.6625,
|
| 1053 |
+
"eval_counterfactual_python_samples_per_second": 0.865,
|
| 1054 |
+
"eval_counterfactual_python_score": -0.10470182448625565,
|
| 1055 |
+
"eval_counterfactual_python_steps_per_second": 0.035,
|
| 1056 |
+
"step": 100
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 1.77,
|
| 1060 |
+
"learning_rate": 0.0002,
|
| 1061 |
+
"loss": 0.2103,
|
| 1062 |
+
"step": 101
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 1.79,
|
| 1066 |
+
"learning_rate": 0.0002,
|
| 1067 |
+
"loss": 0.1066,
|
| 1068 |
+
"step": 102
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 1.81,
|
| 1072 |
+
"learning_rate": 0.0002,
|
| 1073 |
+
"loss": 0.1153,
|
| 1074 |
+
"step": 103
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 1.82,
|
| 1078 |
+
"learning_rate": 0.0002,
|
| 1079 |
+
"loss": 0.1088,
|
| 1080 |
+
"step": 104
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 1.84,
|
| 1084 |
+
"learning_rate": 0.0002,
|
| 1085 |
+
"loss": 0.2025,
|
| 1086 |
+
"step": 105
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 1.86,
|
| 1090 |
+
"learning_rate": 0.0002,
|
| 1091 |
+
"loss": 0.2141,
|
| 1092 |
+
"step": 106
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 1.88,
|
| 1096 |
+
"learning_rate": 0.0002,
|
| 1097 |
+
"loss": 0.1051,
|
| 1098 |
+
"step": 107
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 1.89,
|
| 1102 |
+
"learning_rate": 0.0002,
|
| 1103 |
+
"loss": 0.1626,
|
| 1104 |
+
"step": 108
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 1.91,
|
| 1108 |
+
"learning_rate": 0.0002,
|
| 1109 |
+
"loss": 0.1926,
|
| 1110 |
+
"step": 109
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 1.93,
|
| 1114 |
+
"learning_rate": 0.0002,
|
| 1115 |
+
"loss": 0.2089,
|
| 1116 |
+
"step": 110
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 1.95,
|
| 1120 |
+
"learning_rate": 0.0002,
|
| 1121 |
+
"loss": 0.3994,
|
| 1122 |
+
"step": 111
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 1.96,
|
| 1126 |
+
"learning_rate": 0.0002,
|
| 1127 |
+
"loss": 0.1149,
|
| 1128 |
+
"step": 112
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 1.98,
|
| 1132 |
+
"learning_rate": 0.0002,
|
| 1133 |
+
"loss": 0.195,
|
| 1134 |
+
"step": 113
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 2.0,
|
| 1138 |
+
"learning_rate": 0.0002,
|
| 1139 |
+
"loss": 0.0988,
|
| 1140 |
+
"step": 114
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 2.02,
|
| 1144 |
+
"learning_rate": 0.0002,
|
| 1145 |
+
"loss": 0.0945,
|
| 1146 |
+
"step": 115
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 2.04,
|
| 1150 |
+
"learning_rate": 0.0002,
|
| 1151 |
+
"loss": 0.1407,
|
| 1152 |
+
"step": 116
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 2.05,
|
| 1156 |
+
"learning_rate": 0.0002,
|
| 1157 |
+
"loss": 0.0607,
|
| 1158 |
+
"step": 117
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 2.07,
|
| 1162 |
+
"learning_rate": 0.0002,
|
| 1163 |
+
"loss": 0.0496,
|
| 1164 |
+
"step": 118
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 2.09,
|
| 1168 |
+
"learning_rate": 0.0002,
|
| 1169 |
+
"loss": 0.108,
|
| 1170 |
+
"step": 119
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 2.11,
|
| 1174 |
+
"learning_rate": 0.0002,
|
| 1175 |
+
"loss": 0.0848,
|
| 1176 |
+
"step": 120
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 2.12,
|
| 1180 |
+
"learning_rate": 0.0002,
|
| 1181 |
+
"loss": 0.0976,
|
| 1182 |
+
"step": 121
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 2.14,
|
| 1186 |
+
"learning_rate": 0.0002,
|
| 1187 |
+
"loss": 0.0672,
|
| 1188 |
+
"step": 122
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 2.16,
|
| 1192 |
+
"learning_rate": 0.0002,
|
| 1193 |
+
"loss": 0.0818,
|
| 1194 |
+
"step": 123
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 2.18,
|
| 1198 |
+
"learning_rate": 0.0002,
|
| 1199 |
+
"loss": 0.0695,
|
| 1200 |
+
"step": 124
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 2.19,
|
| 1204 |
+
"learning_rate": 0.0002,
|
| 1205 |
+
"loss": 0.0835,
|
| 1206 |
+
"step": 125
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 2.19,
|
| 1210 |
+
"eval_counterfactual_python_accuracy": 0.89,
|
| 1211 |
+
"eval_counterfactual_python_average_probability": 0.8471024036407471,
|
| 1212 |
+
"eval_counterfactual_python_brier_score": 0.08900060504674911,
|
| 1213 |
+
"eval_counterfactual_python_loss": 0.3583179712295532,
|
| 1214 |
+
"eval_counterfactual_python_probabilities": [
|
| 1215 |
+
0.05809181556105614,
|
| 1216 |
+
0.9953858256340027,
|
| 1217 |
+
0.6866684556007385,
|
| 1218 |
+
1.0,
|
| 1219 |
+
0.9999998807907104,
|
| 1220 |
+
0.9999960660934448,
|
| 1221 |
+
0.9999985694885254,
|
| 1222 |
+
0.9999978542327881,
|
| 1223 |
+
1.0,
|
| 1224 |
+
0.9999997615814209,
|
| 1225 |
+
1.0,
|
| 1226 |
+
1.0,
|
| 1227 |
+
0.990549623966217,
|
| 1228 |
+
0.9999145269393921,
|
| 1229 |
+
0.9818199872970581,
|
| 1230 |
+
0.9946061968803406,
|
| 1231 |
+
0.9380003213882446,
|
| 1232 |
+
0.998694121837616,
|
| 1233 |
+
0.8848497867584229,
|
| 1234 |
+
0.7099964618682861,
|
| 1235 |
+
0.9983460903167725,
|
| 1236 |
+
0.9999994039535522,
|
| 1237 |
+
1.0,
|
| 1238 |
+
1.0,
|
| 1239 |
+
0.49947404861450195,
|
| 1240 |
+
0.5004017949104309,
|
| 1241 |
+
0.49699756503105164,
|
| 1242 |
+
0.9999998807907104,
|
| 1243 |
+
0.9999995231628418,
|
| 1244 |
+
0.9999996423721313,
|
| 1245 |
+
1.0,
|
| 1246 |
+
0.8359315991401672,
|
| 1247 |
+
0.9999997615814209,
|
| 1248 |
+
0.9999977350234985,
|
| 1249 |
+
0.7866907119750977,
|
| 1250 |
+
0.6192131638526917,
|
| 1251 |
+
0.9999843835830688,
|
| 1252 |
+
0.9999998807907104,
|
| 1253 |
+
0.9999995231628418,
|
| 1254 |
+
0.9525272250175476,
|
| 1255 |
+
0.5780937671661377,
|
| 1256 |
+
1.0,
|
| 1257 |
+
0.5239486694335938,
|
| 1258 |
+
0.5134084820747375,
|
| 1259 |
+
0.5220418572425842,
|
| 1260 |
+
0.8747755885124207,
|
| 1261 |
+
0.36531761288642883,
|
| 1262 |
+
0.9437304139137268,
|
| 1263 |
+
0.9974839687347412,
|
| 1264 |
+
0.9675891399383545,
|
| 1265 |
+
0.9971730709075928,
|
| 1266 |
+
0.5026473999023438,
|
| 1267 |
+
0.4968365430831909,
|
| 1268 |
+
0.4916587471961975,
|
| 1269 |
+
0.9146653413772583,
|
| 1270 |
+
0.9551383852958679,
|
| 1271 |
+
0.23804785311222076,
|
| 1272 |
+
1.0,
|
| 1273 |
+
0.9999998807907104,
|
| 1274 |
+
1.0,
|
| 1275 |
+
1.0,
|
| 1276 |
+
1.0,
|
| 1277 |
+
0.9973796606063843,
|
| 1278 |
+
0.29981765151023865,
|
| 1279 |
+
1.0,
|
| 1280 |
+
1.0,
|
| 1281 |
+
0.9998533725738525,
|
| 1282 |
+
0.9999867677688599,
|
| 1283 |
+
0.9904412031173706,
|
| 1284 |
+
0.9999600648880005,
|
| 1285 |
+
0.8782179355621338,
|
| 1286 |
+
0.9403055906295776,
|
| 1287 |
+
0.809188187122345,
|
| 1288 |
+
0.9999572038650513,
|
| 1289 |
+
0.999546468257904,
|
| 1290 |
+
0.5033687353134155,
|
| 1291 |
+
0.511478841304779,
|
| 1292 |
+
0.5243589878082275,
|
| 1293 |
+
0.6943322420120239,
|
| 1294 |
+
0.9999990463256836,
|
| 1295 |
+
0.002750272862613201,
|
| 1296 |
+
0.9997815489768982,
|
| 1297 |
+
0.9998233914375305,
|
| 1298 |
+
0.9999880790710449,
|
| 1299 |
+
0.5942977666854858,
|
| 1300 |
+
1.0,
|
| 1301 |
+
0.9955739974975586,
|
| 1302 |
+
1.0,
|
| 1303 |
+
0.9999998807907104,
|
| 1304 |
+
1.0,
|
| 1305 |
+
0.9997686743736267,
|
| 1306 |
+
0.16937687993049622,
|
| 1307 |
+
0.9986080527305603,
|
| 1308 |
+
0.00019977407646365464,
|
| 1309 |
+
0.9956634640693665,
|
| 1310 |
+
0.995561957359314,
|
| 1311 |
+
0.9999858140945435,
|
| 1312 |
+
0.9999788999557495,
|
| 1313 |
+
0.9999983310699463,
|
| 1314 |
+
0.9999972581863403
|
| 1315 |
+
],
|
| 1316 |
+
"eval_counterfactual_python_runtime": 115.5963,
|
| 1317 |
+
"eval_counterfactual_python_samples_per_second": 0.865,
|
| 1318 |
+
"eval_counterfactual_python_score": -0.08900060504674911,
|
| 1319 |
+
"eval_counterfactual_python_steps_per_second": 0.035,
|
| 1320 |
+
"step": 125
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"loss": 0.1327,
|
| 1324 |
+
"learning_rate": 0.0002,
|
| 1325 |
+
"epoch": 2.21,
|
| 1326 |
+
"step": 126
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"loss": 0.0882,
|
| 1330 |
+
"learning_rate": 0.0002,
|
| 1331 |
+
"epoch": 2.23,
|
| 1332 |
+
"step": 127
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"loss": 0.0875,
|
| 1336 |
+
"learning_rate": 0.0002,
|
| 1337 |
+
"epoch": 2.25,
|
| 1338 |
+
"step": 128
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"loss": 0.0984,
|
| 1342 |
+
"learning_rate": 0.0002,
|
| 1343 |
+
"epoch": 2.26,
|
| 1344 |
+
"step": 129
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"loss": 0.0553,
|
| 1348 |
+
"learning_rate": 0.0002,
|
| 1349 |
+
"epoch": 2.28,
|
| 1350 |
+
"step": 130
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"loss": 0.104,
|
| 1354 |
+
"learning_rate": 0.0002,
|
| 1355 |
+
"epoch": 2.3,
|
| 1356 |
+
"step": 131
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"loss": 0.0297,
|
| 1360 |
+
"learning_rate": 0.0002,
|
| 1361 |
+
"epoch": 2.32,
|
| 1362 |
+
"step": 132
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"loss": 0.0265,
|
| 1366 |
+
"learning_rate": 0.0002,
|
| 1367 |
+
"epoch": 2.33,
|
| 1368 |
+
"step": 133
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"loss": 0.0555,
|
| 1372 |
+
"learning_rate": 0.0002,
|
| 1373 |
+
"epoch": 2.35,
|
| 1374 |
+
"step": 134
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"loss": 0.026,
|
| 1378 |
+
"learning_rate": 0.0002,
|
| 1379 |
+
"epoch": 2.37,
|
| 1380 |
+
"step": 135
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"loss": 0.1426,
|
| 1384 |
+
"learning_rate": 0.0002,
|
| 1385 |
+
"epoch": 2.39,
|
| 1386 |
+
"step": 136
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"loss": 0.102,
|
| 1390 |
+
"learning_rate": 0.0002,
|
| 1391 |
+
"epoch": 2.4,
|
| 1392 |
+
"step": 137
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"loss": 0.0794,
|
| 1396 |
+
"learning_rate": 0.0002,
|
| 1397 |
+
"epoch": 2.42,
|
| 1398 |
+
"step": 138
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"loss": 0.1079,
|
| 1402 |
+
"learning_rate": 0.0002,
|
| 1403 |
+
"epoch": 2.44,
|
| 1404 |
+
"step": 139
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"loss": 0.0256,
|
| 1408 |
+
"learning_rate": 0.0002,
|
| 1409 |
+
"epoch": 2.46,
|
| 1410 |
+
"step": 140
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"loss": 0.048,
|
| 1414 |
+
"learning_rate": 0.0002,
|
| 1415 |
+
"epoch": 2.47,
|
| 1416 |
+
"step": 141
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"loss": 0.0463,
|
| 1420 |
+
"learning_rate": 0.0002,
|
| 1421 |
+
"epoch": 2.49,
|
| 1422 |
+
"step": 142
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"loss": 0.0278,
|
| 1426 |
+
"learning_rate": 0.0002,
|
| 1427 |
+
"epoch": 2.51,
|
| 1428 |
+
"step": 143
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"loss": 0.1124,
|
| 1432 |
+
"learning_rate": 0.0002,
|
| 1433 |
+
"epoch": 2.53,
|
| 1434 |
+
"step": 144
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"loss": 0.0683,
|
| 1438 |
+
"learning_rate": 0.0002,
|
| 1439 |
+
"epoch": 2.54,
|
| 1440 |
+
"step": 145
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"loss": 0.0371,
|
| 1444 |
+
"learning_rate": 0.0002,
|
| 1445 |
+
"epoch": 2.56,
|
| 1446 |
+
"step": 146
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"loss": 0.1198,
|
| 1450 |
+
"learning_rate": 0.0002,
|
| 1451 |
+
"epoch": 2.58,
|
| 1452 |
+
"step": 147
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"loss": 0.0386,
|
| 1456 |
+
"learning_rate": 0.0002,
|
| 1457 |
+
"epoch": 2.6,
|
| 1458 |
+
"step": 148
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"loss": 0.1031,
|
| 1462 |
+
"learning_rate": 0.0002,
|
| 1463 |
+
"epoch": 2.61,
|
| 1464 |
+
"step": 149
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"loss": 0.0731,
|
| 1468 |
+
"learning_rate": 0.0002,
|
| 1469 |
+
"epoch": 2.63,
|
| 1470 |
+
"step": 150
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"eval_counterfactual_python_loss": 0.40711697936058044,
|
| 1474 |
+
"eval_counterfactual_python_score": -0.09575603157281876,
|
| 1475 |
+
"eval_counterfactual_python_brier_score": 0.09575603157281876,
|
| 1476 |
+
"eval_counterfactual_python_average_probability": 0.8392205834388733,
|
| 1477 |
+
"eval_counterfactual_python_accuracy": 0.89,
|
| 1478 |
+
"eval_counterfactual_python_probabilities": [
|
| 1479 |
+
0.2721835970878601,
|
| 1480 |
+
0.9999978542327881,
|
| 1481 |
+
0.5331283211708069,
|
| 1482 |
+
1.0,
|
| 1483 |
+
1.0,
|
| 1484 |
+
1.0,
|
| 1485 |
+
0.9999324083328247,
|
| 1486 |
+
0.9999998807907104,
|
| 1487 |
+
1.0,
|
| 1488 |
+
1.0,
|
| 1489 |
+
1.0,
|
| 1490 |
+
1.0,
|
| 1491 |
+
0.9998410940170288,
|
| 1492 |
+
0.9999990463256836,
|
| 1493 |
+
0.9952380657196045,
|
| 1494 |
+
0.9997170567512512,
|
| 1495 |
+
0.9911946058273315,
|
| 1496 |
+
0.9999902248382568,
|
| 1497 |
+
0.8052052855491638,
|
| 1498 |
+
0.5563430786132812,
|
| 1499 |
+
0.8789305686950684,
|
| 1500 |
+
1.0,
|
| 1501 |
+
1.0,
|
| 1502 |
+
1.0,
|
| 1503 |
+
0.5007253885269165,
|
| 1504 |
+
0.5058515667915344,
|
| 1505 |
+
0.4912094175815582,
|
| 1506 |
+
1.0,
|
| 1507 |
+
1.0,
|
| 1508 |
+
1.0,
|
| 1509 |
+
1.0,
|
| 1510 |
+
0.9610124230384827,
|
| 1511 |
+
1.0,
|
| 1512 |
+
0.9999998807907104,
|
| 1513 |
+
0.7409366369247437,
|
| 1514 |
+
0.5872294306755066,
|
| 1515 |
+
0.9999994039535522,
|
| 1516 |
+
1.0,
|
| 1517 |
+
1.0,
|
| 1518 |
+
0.9307988882064819,
|
| 1519 |
+
0.5262203216552734,
|
| 1520 |
+
1.0,
|
| 1521 |
+
0.6476882100105286,
|
| 1522 |
+
0.5980938076972961,
|
| 1523 |
+
0.6242404580116272,
|
| 1524 |
+
0.6360834836959839,
|
| 1525 |
+
0.2772783637046814,
|
| 1526 |
+
0.9846892356872559,
|
| 1527 |
+
0.9951990246772766,
|
| 1528 |
+
0.9936495423316956,
|
| 1529 |
+
0.999901533126831,
|
| 1530 |
+
0.5019819736480713,
|
| 1531 |
+
0.48769497871398926,
|
| 1532 |
+
0.47629255056381226,
|
| 1533 |
+
0.9037674069404602,
|
| 1534 |
+
0.8574671745300293,
|
| 1535 |
+
0.068470299243927,
|
| 1536 |
+
1.0,
|
| 1537 |
+
1.0,
|
| 1538 |
+
1.0,
|
| 1539 |
+
1.0,
|
| 1540 |
+
1.0,
|
| 1541 |
+
0.9995761513710022,
|
| 1542 |
+
0.24760562181472778,
|
| 1543 |
+
0.9999995231628418,
|
| 1544 |
+
1.0,
|
| 1545 |
+
0.9999412298202515,
|
| 1546 |
+
1.0,
|
| 1547 |
+
0.9950709342956543,
|
| 1548 |
+
0.9999998807907104,
|
| 1549 |
+
0.8830302357673645,
|
| 1550 |
+
0.1665273755788803,
|
| 1551 |
+
0.8684807419776917,
|
| 1552 |
+
0.9999932050704956,
|
| 1553 |
+
0.9999849796295166,
|
| 1554 |
+
0.5084021687507629,
|
| 1555 |
+
0.5264973044395447,
|
| 1556 |
+
0.5384640693664551,
|
| 1557 |
+
0.9106827974319458,
|
| 1558 |
+
1.0,
|
| 1559 |
+
0.00011814858589787036,
|
| 1560 |
+
1.0,
|
| 1561 |
+
1.0,
|
| 1562 |
+
1.0,
|
| 1563 |
+
0.6591076850891113,
|
| 1564 |
+
1.0,
|
| 1565 |
+
0.9994019269943237,
|
| 1566 |
+
1.0,
|
| 1567 |
+
1.0,
|
| 1568 |
+
1.0,
|
| 1569 |
+
1.0,
|
| 1570 |
+
0.2933136820793152,
|
| 1571 |
+
0.9999998807907104,
|
| 1572 |
+
0.00010689133341656998,
|
| 1573 |
+
0.999189555644989,
|
| 1574 |
+
0.9987898468971252,
|
| 1575 |
+
0.9999929666519165,
|
| 1576 |
+
0.9995980858802795,
|
| 1577 |
+
1.0,
|
| 1578 |
+
1.0
|
| 1579 |
+
],
|
| 1580 |
+
"eval_counterfactual_python_runtime": 116.1742,
|
| 1581 |
+
"eval_counterfactual_python_samples_per_second": 0.861,
|
| 1582 |
+
"eval_counterfactual_python_steps_per_second": 0.034,
|
| 1583 |
+
"epoch": 2.63,
|
| 1584 |
+
"step": 150
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"train_runtime": 2419.5441,
|
| 1588 |
+
"train_samples_per_second": 1.984,
|
| 1589 |
+
"train_steps_per_second": 0.062,
|
| 1590 |
+
"total_flos": 0.0,
|
| 1591 |
+
"train_loss": 0.012238509245216847,
|
| 1592 |
+
"epoch": 2.63,
|
| 1593 |
+
"step": 150
|
| 1594 |
+
}
|
| 1595 |
+
]
|