Add files using upload-large-folder tool

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.0.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 645, "1": 0, "2": 0, "3": 0}, "net_accuracy": 0.21371769383697814}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.0.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 645, "1": 0, "2": 0, "3": 0}, "net_accuracy": 0.21371769383697814}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.1.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 0, "1": 0, "2": 783, "3": 0}, "net_accuracy": 0.25944333996023855}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.10.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 0, "1": 483, "2": 237, "3": 47}, "net_accuracy": 0.2541418157720345}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.10.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 645, "1": 0, "2": 0, "3": 0}, "net_accuracy": 0.21371769383697814}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.11.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 5, "1": 0, "2": 30, "3": 833}, "net_accuracy": 0.2876076872100729}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.11.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 645, "1": 0, "2": 0, "3": 0}, "net_accuracy": 0.21371769383697814}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.12.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 86, "1": 608, "2": 5, "3": 0}, "net_accuracy": 0.23161033797216699}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.12.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 645, "1": 0, "2": 0, "3": 0}, "net_accuracy": 0.21371769383697814}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.13.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 0, "1": 0, "2": 783, "3": 0}, "net_accuracy": 0.25944333996023855}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.13.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 645, "1": 0, "2": 0, "3": 0}, "net_accuracy": 0.21371769383697814}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.14.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 2, "1": 625, "2": 111, "3": 1}, "net_accuracy": 0.24486414844267726}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.14.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 630, "1": 3, "2": 0, "3": 7}, "net_accuracy": 0.21206096752816433}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.15.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 65, "1": 2, "2": 1, "3": 796}, "net_accuracy": 0.28628230616302186}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.15.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 645, "1": 0, "2": 0, "3": 1}, "net_accuracy": 0.2140490390987409}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.16.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 335, "1": 243, "2": 30, "3": 38}, "net_accuracy": 0.2140490390987409}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.16.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 550, "1": 122, "2": 0, "3": 0}, "net_accuracy": 0.22266401590457258}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.17.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 264, "1": 77, "2": 17, "3": 405}, "net_accuracy": 0.25281643472498344}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.17.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 641, "1": 4, "2": 0, "3": 0}, "net_accuracy": 0.21371769383697814}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.18.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 66, "1": 10, "2": 0, "3": 748}, "net_accuracy": 0.2730284956925116}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.18.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 176, "1": 1, "2": 0, "3": 592}, "net_accuracy": 0.25480450629555995}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.19.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 553, "1": 0, "2": 0, "3": 142}, "net_accuracy": 0.23028495692511597}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.19.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 35, "1": 0, "2": 0, "3": 830}, "net_accuracy": 0.2866136514247846}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.2.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 0, "1": 0, "2": 0, "3": 871}, "net_accuracy": 0.2886017229953612}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.2.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 522, "1": 0, "2": 0, "3": 187}, "net_accuracy": 0.23492379058979457}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.20.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 0, "1": 19, "2": 747, "3": 12}, "net_accuracy": 0.2577866136514248}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.20.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 106, "1": 0, "2": 0, "3": 718}, "net_accuracy": 0.2730284956925116}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.21.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 237, "1": 0, "2": 466, "3": 0}, "net_accuracy": 0.23293571901921803}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.21.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 558, "1": 0, "2": 0, "3": 129}, "net_accuracy": 0.2276341948310139}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.22.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 639, "1": 0, "2": 9, "3": 0}, "net_accuracy": 0.2147117296222664}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.22.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 585, "1": 0, "2": 0, "3": 96}, "net_accuracy": 0.22564612326043737}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.23.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 640, "1": 7, "2": 0, "3": 0}, "net_accuracy": 0.21438038436050363}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.23.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 622, "1": 0, "2": 0, "3": 37}, "net_accuracy": 0.21835652750165674}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.24.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 607, "1": 0, "2": 34, "3": 13}, "net_accuracy": 0.21669980119284293}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.24.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 637, "1": 0, "2": 0, "3": 17}, "net_accuracy": 0.21669980119284293}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.25.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 122, "1": 597, "2": 10, "3": 1}, "net_accuracy": 0.24188204108681247}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.25.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 636, "1": 0, "2": 0, "3": 17}, "net_accuracy": 0.21636845593108017}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.26.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 1, "1": 135, "2": 0, "3": 691}, "net_accuracy": 0.27402253147779987}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.26.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 621, "1": 0, "2": 0, "3": 39}, "net_accuracy": 0.21868787276341947}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.27.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 0, "1": 43, "2": 559, "3": 176}, "net_accuracy": 0.2577866136514248}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.27.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 578, "1": 0, "2": 0, "3": 102}, "net_accuracy": 0.2253147779986746}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.28.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 613, "1": 3, "2": 26, "3": 5}, "net_accuracy": 0.21438038436050363}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.28.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 640, "1": 0, "2": 0, "3": 9}, "net_accuracy": 0.21504307488402916}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.29.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 146, "1": 256, "2": 374, "3": 0}, "net_accuracy": 0.2571239231278993}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.29.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 643, "1": 0, "2": 0, "3": 3}, "net_accuracy": 0.2140490390987409}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.3.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 0, "1": 713, "2": 3, "3": 5}, "net_accuracy": 0.23889993373094764}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.3.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 645, "1": 0, "2": 0, "3": 0}, "net_accuracy": 0.21371769383697814}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.30.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 0, "1": 0, "2": 783, "3": 0}, "net_accuracy": 0.25944333996023855}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.31.hook_mlp_out.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 0, "1": 717, "2": 0, "3": 0}, "net_accuracy": 0.23757455268389663}

mmlu_STEM/few_shot_0/EleutherAI_gpt-neo-2.7B/last_token_relative_depth_100/accuracy_blocks.31.hook_resid_post.json

@@ -0,0 +1 @@
+{"total_per_class": {"0": 645, "1": 719, "2": 783, "3": 871}, "correct_per_class": {"0": 252, "1": 95, "2": 211, "3": 259}, "net_accuracy": 0.2707090788601723}