lewtun's picture
lewtun HF staff
Upload eval_results/alignment-handbook/zephyr-2b-gemma-sft-v1/main/hellaswag/results_2024-03-05T08-26-29.715216.json with huggingface_hub
a0e78df verified
raw
history blame
2.33 kB
{
"config_general": {
"lighteval_sha": "?",
"num_fewshot_seeds": 1,
"override_batch_size": 1,
"max_samples": null,
"job_id": "",
"start_time": 856411.177187032,
"end_time": 856776.432818228,
"total_evaluation_time_secondes": "365.2556311960798",
"model_name": "alignment-handbook/zephyr-2b-gemma-sft-v1",
"model_sha": "6607539d04bc66d5856772cc04b3b1ce17ebe644",
"model_dtype": "torch.bfloat16",
"model_size": "4.68 GB",
"config": null
},
"results": {
"lighteval|hellaswag|10": {
"acc": 0.5376419040031866,
"acc_stderr": 0.004975621147406098,
"acc_norm": 0.7194781915952997,
"acc_norm_stderr": 0.004483360370140577
}
},
"versions": {
"lighteval|hellaswag|10": 0
},
"config_tasks": {
"lighteval|hellaswag": {
"name": "hellaswag",
"prompt_function": "hellaswag_harness",
"hf_repo": "hellaswag",
"hf_subset": "default",
"metric": [
"loglikelihood_acc",
"loglikelihood_acc_norm"
],
"hf_avail_splits": [
"train",
"test",
"validation"
],
"evaluation_splits": [
"validation"
],
"few_shots_split": null,
"few_shots_select": "random_sampling_from_train",
"generation_size": -1,
"stop_sequence": [
"\n"
],
"output_regex": null,
"frozen": false,
"suite": [
"lighteval"
],
"original_num_docs": 10042,
"effective_num_docs": 10042
}
},
"summary_tasks": {
"lighteval|hellaswag|10": {
"hashes": {
"hash_examples": "31985c805c3a737e",
"hash_full_prompts": "48de8fbcf45195b3",
"hash_input_tokens": "08c832518b7403b1",
"hash_cont_tokens": "6232829aec74d42d"
},
"truncated": 0,
"non_truncated": 10042,
"padded": 39905,
"non_padded": 263,
"effective_few_shots": 10.0,
"num_truncated_few_shots": 0
}
},
"summary_general": {
"hashes": {
"hash_examples": "63bc2cf8bae03fbc",
"hash_full_prompts": "20d63c990d3e8572",
"hash_input_tokens": "e93858038f602554",
"hash_cont_tokens": "c2b17126697982c3"
},
"truncated": 0,
"non_truncated": 10042,
"padded": 39905,
"non_padded": 263,
"num_truncated_few_shots": 0
}
}