longformer-qasper-document-qa / evaluation /controlled_context_length_comparison.json
anmol-unitmole's picture
Upload evaluation/controlled_context_length_comparison.json with huggingface_hub
b43aa89 verified
Raw
History Blame Contribute Delete
5.5 kB
[
{
"model_name": "BERT truncated 512",
"controlled_target_tokens": 384,
"examples": 12,
"exact_match": 0.4166666666666667,
"token_f1": 0.5696649029982364,
"evidence_recovered": 0.8333333333333334,
"evidence_token_recall": 0.9037037037037038,
"average_latency_seconds": 0.020591975000570528,
"average_windows": 1.0
},
{
"model_name": "BERT truncated 512",
"controlled_target_tokens": 768,
"examples": 12,
"exact_match": 0.08333333333333333,
"token_f1": 0.08950617283950617,
"evidence_recovered": 0.16666666666666666,
"evidence_token_recall": 0.27483828629114954,
"average_latency_seconds": 0.02036956666658322,
"average_windows": 1.0
},
{
"model_name": "BERT truncated 512",
"controlled_target_tokens": 1536,
"examples": 12,
"exact_match": 0.08333333333333333,
"token_f1": 0.14773763911694945,
"evidence_recovered": 0.16666666666666666,
"evidence_token_recall": 0.3438101394225293,
"average_latency_seconds": 0.02243222499479695,
"average_windows": 1.0
},
{
"model_name": "BERT truncated 512",
"controlled_target_tokens": 3072,
"examples": 12,
"exact_match": 0.08333333333333333,
"token_f1": 0.10275835275835277,
"evidence_recovered": 0.08333333333333333,
"evidence_token_recall": 0.2376230463147804,
"average_latency_seconds": 0.025531658340090264,
"average_windows": 1.0
},
{
"model_name": "BERT truncated 512",
"controlled_target_tokens": 4608,
"examples": 12,
"exact_match": 0.08333333333333333,
"token_f1": 0.1048951048951049,
"evidence_recovered": 0.08333333333333333,
"evidence_token_recall": 0.2992065754834572,
"average_latency_seconds": 0.028508583335982014,
"average_windows": 1.0
},
{
"model_name": "Longformer SQuAD sliding windows",
"controlled_target_tokens": 384,
"examples": 12,
"exact_match": 0.3333333333333333,
"token_f1": 0.4934343434343434,
"evidence_recovered": 0.75,
"evidence_token_recall": 0.8358024691358025,
"average_latency_seconds": 0.06816069167204357,
"average_windows": 1.0
},
{
"model_name": "Longformer SQuAD sliding windows",
"controlled_target_tokens": 768,
"examples": 12,
"exact_match": 0.25,
"token_f1": 0.4136738906088751,
"evidence_recovered": 0.5833333333333334,
"evidence_token_recall": 0.729773994424532,
"average_latency_seconds": 0.07133218333668385,
"average_windows": 1.0
},
{
"model_name": "Longformer SQuAD sliding windows",
"controlled_target_tokens": 1536,
"examples": 12,
"exact_match": 0.0,
"token_f1": 0.13654970760233917,
"evidence_recovered": 0.25,
"evidence_token_recall": 0.45772529887152297,
"average_latency_seconds": 0.08053487500001211,
"average_windows": 1.0
},
{
"model_name": "Longformer SQuAD sliding windows",
"controlled_target_tokens": 3072,
"examples": 12,
"exact_match": 0.16666666666666666,
"token_f1": 0.3242424242424242,
"evidence_recovered": 0.4166666666666667,
"evidence_token_recall": 0.5802561217227383,
"average_latency_seconds": 0.15887252500397153,
"average_windows": 2.0
},
{
"model_name": "Longformer SQuAD sliding windows",
"controlled_target_tokens": 4608,
"examples": 12,
"exact_match": 0.08333333333333333,
"token_f1": 0.1414141414141414,
"evidence_recovered": 0.25,
"evidence_token_recall": 0.4190843997467528,
"average_latency_seconds": 0.23718954166785503,
"average_windows": 3.0
},
{
"model_name": "Longformer QASPER fine-tuned",
"controlled_target_tokens": 384,
"examples": 12,
"exact_match": 0.5,
"token_f1": 0.6670454545454545,
"evidence_recovered": 0.8333333333333334,
"evidence_token_recall": 0.9037037037037038,
"average_latency_seconds": 0.07243051666591782,
"average_windows": 1.0
},
{
"model_name": "Longformer QASPER fine-tuned",
"controlled_target_tokens": 768,
"examples": 12,
"exact_match": 0.5,
"token_f1": 0.6670454545454545,
"evidence_recovered": 0.8333333333333334,
"evidence_token_recall": 0.9037037037037038,
"average_latency_seconds": 0.07527149165495454,
"average_windows": 1.0
},
{
"model_name": "Longformer QASPER fine-tuned",
"controlled_target_tokens": 1536,
"examples": 12,
"exact_match": 0.25,
"token_f1": 0.42843434343434345,
"evidence_recovered": 0.6666666666666666,
"evidence_token_recall": 0.7766862170087977,
"average_latency_seconds": 0.07720827500937351,
"average_windows": 1.0
},
{
"model_name": "Longformer QASPER fine-tuned",
"controlled_target_tokens": 3072,
"examples": 12,
"exact_match": 0.25,
"token_f1": 0.4398148148148148,
"evidence_recovered": 0.6666666666666666,
"evidence_token_recall": 0.7799663299663299,
"average_latency_seconds": 0.16716532500868198,
"average_windows": 2.0
},
{
"model_name": "Longformer QASPER fine-tuned",
"controlled_target_tokens": 4608,
"examples": 12,
"exact_match": 0.25,
"token_f1": 0.4398148148148148,
"evidence_recovered": 0.6666666666666666,
"evidence_token_recall": 0.7818602693602693,
"average_latency_seconds": 0.2461085750013202,
"average_windows": 3.0
}
]