{ "avg_r2": 0.9650599151719406, "avg_mae": 0.00154096947517246, "thresholds_used": { "small": 412876, "large": 1651507 }, "r2_layers": [ { "layer_name": "bert.embeddings.word_embeddings.weight", "r2_score": 0.9998854142946492, "num_samples": 16408627 }, { "layer_name": "bert.embeddings.position_embeddings.weight", "r2_score": 0.9795404869240352, "num_samples": 275251 }, { "layer_name": "bert.embeddings.token_type_embeddings.weight", "r2_score": -0.26726293685598956, "num_samples": 1075 }, { "layer_name": "bert.encoder.layer.0.attention.self.query.weight", "r2_score": 0.996277390558186, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.0.attention.self.key.weight", "r2_score": 0.9956420978468314, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.0.attention.self.value.weight", "r2_score": 0.9967422690651456, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.0.attention.output.dense.weight", "r2_score": 0.9974010598893565, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.0.intermediate.dense.weight", "r2_score": 0.9975661672133151, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.0.output.dense.weight", "r2_score": 0.9983455569568541, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.1.attention.self.query.weight", "r2_score": 0.9939746532909586, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.1.attention.self.key.weight", "r2_score": 0.9969038409594635, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.1.attention.self.value.weight", "r2_score": 0.9963270362399188, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.1.attention.output.dense.weight", "r2_score": 0.993517732174246, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.1.intermediate.dense.weight", "r2_score": 0.9985278305910291, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.1.output.dense.weight", "r2_score": 0.998033121130939, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.2.attention.self.query.weight", "r2_score": 0.9935661795048506, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.2.attention.self.key.weight", "r2_score": 0.9944536144836053, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.2.attention.self.value.weight", "r2_score": 0.9966468839292184, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.2.attention.output.dense.weight", "r2_score": 0.9960558417729032, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.2.intermediate.dense.weight", "r2_score": 0.9980592016826356, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.2.output.dense.weight", "r2_score": 0.9975617459868251, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.3.attention.self.query.weight", "r2_score": 0.995230398629041, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.3.attention.self.key.weight", "r2_score": 0.9933370340154851, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.3.attention.self.value.weight", "r2_score": 0.9958764790905029, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.3.attention.output.dense.weight", "r2_score": 0.99078311378124, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.3.intermediate.dense.weight", "r2_score": 0.9984591962092705, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.3.output.dense.weight", "r2_score": 0.9981946928220499, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.4.attention.self.query.weight", "r2_score": 0.9973075711345991, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.4.attention.self.key.weight", "r2_score": 0.9977659790811637, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.4.attention.self.value.weight", "r2_score": 0.9948040198297013, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.4.attention.output.dense.weight", "r2_score": 0.9956406281779216, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.4.intermediate.dense.weight", "r2_score": 0.9987461663239621, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.4.output.dense.weight", "r2_score": 0.9983982323964519, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.5.attention.self.query.weight", "r2_score": 0.9967685797161927, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.5.attention.self.key.weight", "r2_score": 0.994687568099075, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.5.attention.self.value.weight", "r2_score": 0.9951222636213912, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.5.attention.output.dense.weight", "r2_score": 0.9965343395951053, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.5.intermediate.dense.weight", "r2_score": 0.9979230597638107, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.5.output.dense.weight", "r2_score": 0.9987543460214953, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.6.attention.self.query.weight", "r2_score": 0.9941675902331104, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.6.attention.self.key.weight", "r2_score": 0.9972271864613587, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.6.attention.self.value.weight", "r2_score": 0.99611595659062, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.6.attention.output.dense.weight", "r2_score": 0.9960385358549926, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.6.intermediate.dense.weight", "r2_score": 0.9977964593420854, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.6.output.dense.weight", "r2_score": 0.9971090954347793, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.7.attention.self.query.weight", "r2_score": 0.9962744123031244, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.7.attention.self.key.weight", "r2_score": 0.9965041999319837, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.7.attention.self.value.weight", "r2_score": 0.994934947624443, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.7.attention.output.dense.weight", "r2_score": 0.9956120664155155, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.7.intermediate.dense.weight", "r2_score": 0.9976038292034548, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.7.output.dense.weight", "r2_score": 0.9972641644624681, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.8.attention.self.query.weight", "r2_score": 0.9950410538090237, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.8.attention.self.key.weight", "r2_score": 0.9966817641105654, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.8.attention.self.value.weight", "r2_score": 0.9963506682062593, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.8.attention.output.dense.weight", "r2_score": 0.9952435682482271, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.8.intermediate.dense.weight", "r2_score": 0.9977305487510404, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.8.output.dense.weight", "r2_score": 0.9970859861549761, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.9.attention.self.query.weight", "r2_score": 0.9964727133202562, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.9.attention.self.key.weight", "r2_score": 0.9954619045086586, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.9.attention.self.value.weight", "r2_score": 0.9941545352691554, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.9.attention.output.dense.weight", "r2_score": 0.9964365278215438, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.9.intermediate.dense.weight", "r2_score": 0.9981828306300096, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.9.output.dense.weight", "r2_score": 0.9978749831277458, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.10.attention.self.query.weight", "r2_score": 0.9954721855513622, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.10.attention.self.key.weight", "r2_score": 0.9967209618336382, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.10.attention.self.value.weight", "r2_score": 0.9969923492158588, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.10.attention.output.dense.weight", "r2_score": 0.9966273316514799, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.10.intermediate.dense.weight", "r2_score": 0.9986453321100079, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.10.output.dense.weight", "r2_score": 0.997803539947793, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.11.attention.self.query.weight", "r2_score": 0.9944997930283149, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.11.attention.self.key.weight", "r2_score": 0.9963109082402576, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.11.attention.self.value.weight", "r2_score": 0.9948018697403358, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.11.attention.output.dense.weight", "r2_score": 0.9955671566641768, "num_samples": 412876 }, { "layer_name": "bert.encoder.layer.11.intermediate.dense.weight", "r2_score": 0.9980132278783622, "num_samples": 1651507 }, { "layer_name": "bert.encoder.layer.11.output.dense.weight", "r2_score": 0.9981850721917447, "num_samples": 1651507 }, { "layer_name": "bert.pooler.dense.weight", "r2_score": 0.9959159457910624, "num_samples": 412876 }, { "layer_name": "classifier.weight", "r2_score": -0.1414065893678227, "num_samples": 1075 } ] }