Hugging Face's logo

Datasets:
Exploration-Lab
/
dim-discovery-archive

ArXiv:
Dataset card Files
xet
 Community
Divyanshu-Bhatt commited on
Commit
f955fe1
·
verified ·
1 Parent(s): 70d2d77

Add files using upload-large-folder tool

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.0.hook_mlp_out.json +1 -0
  2. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.1.hook_mlp_out.json +1 -0
  3. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.10.hook_mlp_out.json +1 -0
  4. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.11.hook_mlp_out.json +1 -0
  5. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.12.hook_mlp_out.json +1 -0
  6. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.12.hook_resid_post.json +1 -0
  7. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.13.hook_mlp_out.json +1 -0
  8. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.13.hook_resid_post.json +1 -0
  9. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.14.hook_mlp_out.json +1 -0
  10. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.15.hook_mlp_out.json +1 -0
  11. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.16.hook_mlp_out.json +1 -0
  12. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.16.hook_resid_post.json +1 -0
  13. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.17.hook_mlp_out.json +1 -0
  14. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.17.hook_resid_post.json +1 -0
  15. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.18.hook_resid_post.json +1 -0
  16. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.19.hook_mlp_out.json +1 -0
  17. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.19.hook_resid_post.json +1 -0
  18. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.2.hook_mlp_out.json +1 -0
  19. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.2.hook_resid_post.json +1 -0
  20. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.20.hook_mlp_out.json +1 -0
  21. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.20.hook_resid_post.json +1 -0
  22. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.21.hook_mlp_out.json +1 -0
  23. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.21.hook_resid_post.json +1 -0
  24. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.22.hook_mlp_out.json +1 -0
  25. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.23.hook_mlp_out.json +1 -0
  26. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.23.hook_resid_post.json +1 -0
  27. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.24.hook_mlp_out.json +1 -0
  28. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.24.hook_resid_post.json +1 -0
  29. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.25.hook_mlp_out.json +1 -0
  30. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.25.hook_resid_post.json +1 -0
  31. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.26.hook_mlp_out.json +1 -0
  32. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.26.hook_resid_post.json +1 -0
  33. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.27.hook_mlp_out.json +1 -0
  34. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.28.hook_mlp_out.json +1 -0
  35. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.29.hook_resid_post.json +1 -0
  36. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.3.hook_mlp_out.json +1 -0
  37. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.3.hook_resid_post.json +1 -0
  38. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.31.hook_mlp_out.json +1 -0
  39. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.4.hook_mlp_out.json +1 -0
  40. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.5.hook_mlp_out.json +1 -0
  41. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.6.hook_mlp_out.json +1 -0
  42. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.6.hook_resid_post.json +1 -0
  43. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.7.hook_mlp_out.json +1 -0
  44. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.7.hook_resid_post.json +1 -0
  45. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.8.hook_mlp_out.json +1 -0
  46. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.8.hook_resid_post.json +1 -0
  47. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.9.hook_resid_post.json +1 -0
  48. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_ln_final.hook_normalized.json +1 -0
  49. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_output.json +1 -0
  50. copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/intrinsic/mle_local_dims/blocks.0.hook_mlp_out.csv +0 -0
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.0.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 0, "1": 511}, "net_accuracy": 0.511}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.1.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 0, "1": 511}, "net_accuracy": 0.511}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.10.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 0, "1": 511}, "net_accuracy": 0.511}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.11.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 424, "1": 69}, "net_accuracy": 0.493}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.12.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 358, "1": 110}, "net_accuracy": 0.468}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.12.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 328, "1": 152}, "net_accuracy": 0.48}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.13.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 375, "1": 156}, "net_accuracy": 0.531}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.13.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 432, "1": 9}, "net_accuracy": 0.441}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.14.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 189, "1": 382}, "net_accuracy": 0.571}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.15.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 2, "1": 504}, "net_accuracy": 0.506}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.16.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 112, "1": 357}, "net_accuracy": 0.469}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.16.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 361, "1": 365}, "net_accuracy": 0.726}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.17.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 474, "1": 3}, "net_accuracy": 0.477}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.17.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 430, "1": 436}, "net_accuracy": 0.866}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.18.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 430, "1": 430}, "net_accuracy": 0.86}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.19.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 34, "1": 500}, "net_accuracy": 0.534}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.19.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 435, "1": 429}, "net_accuracy": 0.864}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.2.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 489, "1": 0}, "net_accuracy": 0.489}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.2.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 0, "1": 511}, "net_accuracy": 0.511}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.20.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 223, "1": 476}, "net_accuracy": 0.699}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.20.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 435, "1": 429}, "net_accuracy": 0.864}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.21.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 0, "1": 509}, "net_accuracy": 0.509}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.21.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 426, "1": 440}, "net_accuracy": 0.866}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.22.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 422, "1": 434}, "net_accuracy": 0.856}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.23.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 71, "1": 75}, "net_accuracy": 0.146}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.23.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 422, "1": 451}, "net_accuracy": 0.873}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.24.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 91, "1": 53}, "net_accuracy": 0.144}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.24.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 200, "1": 503}, "net_accuracy": 0.703}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.25.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 413, "1": 370}, "net_accuracy": 0.783}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.25.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 367, "1": 477}, "net_accuracy": 0.844}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.26.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 427, "1": 434}, "net_accuracy": 0.861}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.26.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 405, "1": 468}, "net_accuracy": 0.873}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.27.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 300, "1": 493}, "net_accuracy": 0.793}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.28.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 378, "1": 480}, "net_accuracy": 0.858}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.29.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 407, "1": 464}, "net_accuracy": 0.871}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.3.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 311, "1": 178}, "net_accuracy": 0.489}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.3.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 488, "1": 2}, "net_accuracy": 0.49}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.31.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 68, "1": 86}, "net_accuracy": 0.154}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.4.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 489, "1": 0}, "net_accuracy": 0.489}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.5.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 188, "1": 299}, "net_accuracy": 0.487}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.6.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 489, "1": 0}, "net_accuracy": 0.489}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.6.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 489, "1": 0}, "net_accuracy": 0.489}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.7.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 0, "1": 511}, "net_accuracy": 0.511}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.7.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 489, "1": 0}, "net_accuracy": 0.489}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.8.hook_mlp_out.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 489, "1": 0}, "net_accuracy": 0.489}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.8.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 489, "1": 0}, "net_accuracy": 0.489}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_blocks.9.hook_resid_post.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 489, "1": 0}, "net_accuracy": 0.489}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_ln_final.hook_normalized.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 416, "1": 455}, "net_accuracy": 0.871}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/accuracy_output.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"total_per_class": {"0": 489, "1": 511}, "correct_per_class": {"0": 416, "1": 455}, "net_accuracy": 0.871}
copa_train/few_shot_5/meta-llama_Meta-Llama-3-8B-Instruct/last_token_relative_depth_100/intrinsic/mle_local_dims/blocks.0.hook_mlp_out.csv ADDED
The diff for this file is too large to render. See raw diff