lewtun's picture
lewtun HF staff
Upload eval_results/openchat/openchat-3.5-0106/main/arc/results_2024-03-02T18-58-10.340831.json with huggingface_hub
d62e492 verified
raw
history blame
2.31 kB
{
"config_general": {
"lighteval_sha": "?",
"num_fewshot_seeds": 1,
"override_batch_size": 1,
"max_samples": null,
"job_id": "",
"start_time": 741670.256612102,
"end_time": 741808.533565316,
"total_evaluation_time_secondes": "138.2769532139646",
"model_name": "openchat/openchat-3.5-0106",
"model_sha": "9619fb7d2a8e25fa6b0633c0f57f7f4aa79b45c4",
"model_dtype": "torch.bfloat16",
"model_size": "13.61 GB",
"config": null
},
"results": {
"lighteval|arc:challenge|25": {
"acc": 0.5981228668941979,
"acc_stderr": 0.014327268614578274,
"acc_norm": 0.6100682593856656,
"acc_norm_stderr": 0.014252959848892901
}
},
"versions": {
"lighteval|arc:challenge|25": 0
},
"config_tasks": {
"lighteval|arc:challenge": {
"name": "arc:challenge",
"prompt_function": "arc",
"hf_repo": "ai2_arc",
"hf_subset": "ARC-Challenge",
"metric": [
"loglikelihood_acc",
"loglikelihood_acc_norm_nospace"
],
"hf_avail_splits": [
"train",
"test"
],
"evaluation_splits": [
"test"
],
"few_shots_split": null,
"few_shots_select": "random_sampling_from_train",
"generation_size": 1,
"stop_sequence": [
"\n"
],
"output_regex": null,
"frozen": false,
"suite": [
"lighteval",
"arc"
],
"original_num_docs": 1172,
"effective_num_docs": 1172
}
},
"summary_tasks": {
"lighteval|arc:challenge|25": {
"hashes": {
"hash_examples": "17b0cae357c0259e",
"hash_full_prompts": "0f771296c16023aa",
"hash_input_tokens": "cd2659834ac831f8",
"hash_cont_tokens": "289aa98c400841d8"
},
"truncated": 0,
"non_truncated": 1172,
"padded": 4676,
"non_padded": 11,
"effective_few_shots": 25.0,
"num_truncated_few_shots": 0
}
},
"summary_general": {
"hashes": {
"hash_examples": "aaa6929c6d3771fb",
"hash_full_prompts": "2f4738dda6824b30",
"hash_input_tokens": "a6532e8ba3b463ed",
"hash_cont_tokens": "a2928b768b852bf8"
},
"truncated": 0,
"non_truncated": 1172,
"padded": 4676,
"non_padded": 11,
"num_truncated_few_shots": 0
}
}