Add new models

102/edges.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a2cb02e0518a50c2b53be59ea3c3788ec80c0d3c6c1e70070bd7115abdfae77
+size 113

102/ll_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:feb81e3d0658fbd63f8e19b5ec8b036dd18f5e58ef62d8e0a57048e6ca450c7d
+size 15082

102/ll_model_cfg.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ecb69a2f8949b8c7b19adc4c33aed4e99913687a72b169b64a317d2e7878dc97
+size 1093

102/meta.json

@@ -0,0 +1 @@
+{"output_dir": "/circuits-benchmark/results", "atol": 0.05, "lr": 0.001, "use_single_loss": true, "iit_weight": 1.0, "behavior_weight": 0.4, "strict_weight": 0.4, "epochs": 1000, "early_stop_accuracy_threshold": 99.9, "act_fn": "gelu", "use_wandb": true, "wandb_project": "iit-train-seed-82-siit-weigth-0.4", "wandb_name": "case-102-seed-82-s-0.4-b-0.4-iit-1", "save_model_to_wandb": true, "device": "cuda", "clip_grad_norm": 0.1, "lr_scheduler": "linear", "model_pair": "strict", "same_size": false, "seed": 82, "batch_size": 256, "include_mlp": false, "detach_while_caching": true, "scheduler_val_metric": ["val/accuracy", "val/IIA", "val/strict_accuracy"], "siit_sampling": "sample_all", "val_iia_sampling": "all"}

104/edges.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a2cb02e0518a50c2b53be59ea3c3788ec80c0d3c6c1e70070bd7115abdfae77
+size 113

104/ll_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddb4db579c13e1bf510ead6a01c343582eedff2f1d799f4d733d72f5f9da50ab
+size 14698

104/ll_model_cfg.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db76917a80935813594508b1d79b7a0836a20daf67e263161432430b095fd111
+size 1093

104/meta.json

@@ -0,0 +1 @@
+{"output_dir": "/circuits-benchmark/results", "atol": 0.05, "lr": 0.001, "use_single_loss": true, "iit_weight": 1.0, "behavior_weight": 0.4, "strict_weight": 0.7, "epochs": 1000, "early_stop_accuracy_threshold": 99.9, "act_fn": "gelu", "use_wandb": true, "wandb_project": "iit-train-seed-67-siit-weigth-0.7", "wandb_name": "case-104-seed-67-s-0.7-b-0.4-iit-1", "save_model_to_wandb": true, "device": "cuda", "clip_grad_norm": 0.1, "lr_scheduler": "linear", "model_pair": "strict", "same_size": false, "seed": 67, "batch_size": 256, "include_mlp": false, "detach_while_caching": true, "scheduler_val_metric": ["val/accuracy", "val/IIA", "val/strict_accuracy"], "siit_sampling": "sample_all", "val_iia_sampling": "all"}

105/edges.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a2cb02e0518a50c2b53be59ea3c3788ec80c0d3c6c1e70070bd7115abdfae77
+size 113

105/ll_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c32e3e97b224cc8d115b70d7661b8b8433bfcd41a9f58d1d7409ce9367ea27a
+size 15018

105/ll_model_cfg.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:57dece33450716de87f38b12c0955174849706456cc698dd4fcce152feba3cbf
+size 1093

105/meta.json

@@ -0,0 +1 @@
+{"output_dir": "/circuits-benchmark/results", "atol": 0.05, "lr": 0.001, "use_single_loss": true, "iit_weight": 1.0, "behavior_weight": 0.4, "strict_weight": 0.7, "epochs": 1000, "early_stop_accuracy_threshold": 99.9, "act_fn": "gelu", "use_wandb": true, "wandb_project": "iit-train-seed-67-siit-weigth-0.7", "wandb_name": "case-105-seed-67-s-0.7-b-0.4-iit-1", "save_model_to_wandb": true, "device": "cuda", "clip_grad_norm": 0.1, "lr_scheduler": "linear", "model_pair": "strict", "same_size": false, "seed": 67, "batch_size": 256, "include_mlp": false, "detach_while_caching": true, "scheduler_val_metric": ["val/accuracy", "val/IIA", "val/strict_accuracy"], "siit_sampling": "sample_all", "val_iia_sampling": "all"}

123/edges.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a2cb02e0518a50c2b53be59ea3c3788ec80c0d3c6c1e70070bd7115abdfae77
+size 113

123/ll_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b640bb628be87f8ebf41e6bb8b5351a6d69bfa11699bf2e1cea140f6f6f9f95
+size 15082

123/ll_model_cfg.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f027ad4de6828c6f5bb7a3f3c8aabd9658526e33b2284eba366977823b89c0a
+size 1093

123/meta.json

@@ -0,0 +1 @@
+{"output_dir": "/circuits-benchmark/results", "atol": 0.05, "lr": 0.001, "use_single_loss": true, "iit_weight": 1.0, "behavior_weight": 0.4, "strict_weight": 0.7, "epochs": 1000, "early_stop_accuracy_threshold": 99.9, "act_fn": "gelu", "use_wandb": true, "wandb_project": "iit-train-seed-67-siit-weigth-0.7", "wandb_name": "case-123-seed-67-s-0.7-b-0.4-iit-1", "save_model_to_wandb": true, "device": "cuda", "clip_grad_norm": 0.1, "lr_scheduler": "linear", "model_pair": "strict", "same_size": false, "seed": 67, "batch_size": 256, "include_mlp": false, "detach_while_caching": true, "scheduler_val_metric": ["val/accuracy", "val/IIA", "val/strict_accuracy"], "siit_sampling": "sample_all", "val_iia_sampling": "all"}

46/edges.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a2cb02e0518a50c2b53be59ea3c3788ec80c0d3c6c1e70070bd7115abdfae77
+size 113

46/ll_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddb4db579c13e1bf510ead6a01c343582eedff2f1d799f4d733d72f5f9da50ab
+size 14698

46/ll_model_cfg.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db76917a80935813594508b1d79b7a0836a20daf67e263161432430b095fd111
+size 1093

46/meta.json

@@ -0,0 +1 @@
+{"output_dir": "/circuits-benchmark/results", "atol": 0.05, "lr": 0.001, "use_single_loss": true, "iit_weight": 1.0, "behavior_weight": 0.4, "strict_weight": 0.7, "epochs": 1000, "early_stop_accuracy_threshold": 99.9, "act_fn": "gelu", "use_wandb": true, "wandb_project": "iit-train-seed-67-siit-weigth-0.7", "wandb_name": "case-46-seed-67-s-0.7-b-0.4-iit-1", "save_model_to_wandb": true, "device": "cuda", "clip_grad_norm": 0.1, "lr_scheduler": "linear", "model_pair": "strict", "same_size": false, "seed": 67, "batch_size": 256, "include_mlp": false, "detach_while_caching": true, "scheduler_val_metric": ["val/accuracy", "val/IIA", "val/strict_accuracy"], "siit_sampling": "sample_all", "val_iia_sampling": "all"}

50/edges.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a2cb02e0518a50c2b53be59ea3c3788ec80c0d3c6c1e70070bd7115abdfae77
+size 113

50/ll_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddb4db579c13e1bf510ead6a01c343582eedff2f1d799f4d733d72f5f9da50ab
+size 14698

50/ll_model_cfg.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db76917a80935813594508b1d79b7a0836a20daf67e263161432430b095fd111
+size 1093

50/meta.json

@@ -0,0 +1 @@
+{"output_dir": "/circuits-benchmark/results", "atol": 0.05, "lr": 0.001, "use_single_loss": true, "iit_weight": 1.0, "behavior_weight": 0.4, "strict_weight": 0.7, "epochs": 1000, "early_stop_accuracy_threshold": 99.9, "act_fn": "gelu", "use_wandb": true, "wandb_project": "iit-train-seed-67-siit-weigth-0.7", "wandb_name": "case-50-seed-67-s-0.7-b-0.4-iit-1", "save_model_to_wandb": true, "device": "cuda", "clip_grad_norm": 0.1, "lr_scheduler": "linear", "model_pair": "strict", "same_size": false, "seed": 67, "batch_size": 256, "include_mlp": false, "detach_while_caching": true, "scheduler_val_metric": ["val/accuracy", "val/IIA", "val/strict_accuracy"], "siit_sampling": "sample_all", "val_iia_sampling": "all"}

52/edges.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a2cb02e0518a50c2b53be59ea3c3788ec80c0d3c6c1e70070bd7115abdfae77
+size 113

52/ll_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddb4db579c13e1bf510ead6a01c343582eedff2f1d799f4d733d72f5f9da50ab
+size 14698

52/ll_model_cfg.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db76917a80935813594508b1d79b7a0836a20daf67e263161432430b095fd111
+size 1093

52/meta.json

@@ -0,0 +1 @@
+{"output_dir": "/circuits-benchmark/results", "atol": 0.05, "lr": 0.001, "use_single_loss": true, "iit_weight": 1.0, "behavior_weight": 0.4, "strict_weight": 0.7, "epochs": 1000, "early_stop_accuracy_threshold": 99.9, "act_fn": "gelu", "use_wandb": true, "wandb_project": "iit-train-seed-67-siit-weigth-0.7", "wandb_name": "case-52-seed-67-s-0.7-b-0.4-iit-1", "save_model_to_wandb": true, "device": "cuda", "clip_grad_norm": 0.1, "lr_scheduler": "linear", "model_pair": "strict", "same_size": false, "seed": 67, "batch_size": 256, "include_mlp": false, "detach_while_caching": true, "scheduler_val_metric": ["val/accuracy", "val/IIA", "val/strict_accuracy"], "siit_sampling": "sample_all", "val_iia_sampling": "all"}

53/edges.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:994f3bdfbe5b148e1da38018a24a1567c1d86e5de1c18e9b4d62af358812c709
+size 189

53/ll_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c16ef496b74897cf166795daa3a019dd6adde99efe3edd12d540676047ff695b
+size 14762

53/ll_model_cfg.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:067caa3354f7d42fd2eb39aeacec1719e1e8aa60f9c707b5be1dff2a7a5eac4c
+size 1093

53/meta.json

@@ -0,0 +1 @@
+{"output_dir": "/circuits-benchmark/results", "atol": 0.05, "lr": 0.001, "use_single_loss": true, "iit_weight": 1.0, "behavior_weight": 0.4, "strict_weight": 0.7, "epochs": 1000, "early_stop_accuracy_threshold": 99.9, "act_fn": "gelu", "use_wandb": true, "wandb_project": "iit-train-seed-67-siit-weigth-0.7", "wandb_name": "case-53-seed-67-s-0.7-b-0.4-iit-1", "save_model_to_wandb": true, "device": "cuda", "clip_grad_norm": 0.1, "lr_scheduler": "linear", "model_pair": "strict", "same_size": false, "seed": 67, "batch_size": 256, "include_mlp": false, "detach_while_caching": true, "scheduler_val_metric": ["val/accuracy", "val/IIA", "val/strict_accuracy"], "siit_sampling": "sample_all", "val_iia_sampling": "all"}

54/edges.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a2cb02e0518a50c2b53be59ea3c3788ec80c0d3c6c1e70070bd7115abdfae77
+size 113

54/ll_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddb4db579c13e1bf510ead6a01c343582eedff2f1d799f4d733d72f5f9da50ab
+size 14698

54/ll_model_cfg.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db76917a80935813594508b1d79b7a0836a20daf67e263161432430b095fd111
+size 1093

54/meta.json

@@ -0,0 +1 @@
+{"output_dir": "/circuits-benchmark/results", "atol": 0.05, "lr": 0.001, "use_single_loss": true, "iit_weight": 1.0, "behavior_weight": 0.4, "strict_weight": 0.7, "epochs": 1000, "early_stop_accuracy_threshold": 99.9, "act_fn": "gelu", "use_wandb": true, "wandb_project": "iit-train-seed-67-siit-weigth-0.7", "wandb_name": "case-54-seed-67-s-0.7-b-0.4-iit-1", "save_model_to_wandb": true, "device": "cuda", "clip_grad_norm": 0.1, "lr_scheduler": "linear", "model_pair": "strict", "same_size": false, "seed": 67, "batch_size": 256, "include_mlp": false, "detach_while_caching": true, "scheduler_val_metric": ["val/accuracy", "val/IIA", "val/strict_accuracy"], "siit_sampling": "sample_all", "val_iia_sampling": "all"}

55/edges.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a2cb02e0518a50c2b53be59ea3c3788ec80c0d3c6c1e70070bd7115abdfae77
+size 113

55/ll_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddb4db579c13e1bf510ead6a01c343582eedff2f1d799f4d733d72f5f9da50ab
+size 14698

55/ll_model_cfg.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db76917a80935813594508b1d79b7a0836a20daf67e263161432430b095fd111
+size 1093

55/meta.json

@@ -0,0 +1 @@
+{"output_dir": "/circuits-benchmark/results", "atol": 0.05, "lr": 0.001, "use_single_loss": true, "iit_weight": 1.0, "behavior_weight": 0.4, "strict_weight": 0.7, "epochs": 1000, "early_stop_accuracy_threshold": 99.9, "act_fn": "gelu", "use_wandb": true, "wandb_project": "iit-train-seed-67-siit-weigth-0.7", "wandb_name": "case-55-seed-67-s-0.7-b-0.4-iit-1", "save_model_to_wandb": true, "device": "cuda", "clip_grad_norm": 0.1, "lr_scheduler": "linear", "model_pair": "strict", "same_size": false, "seed": 67, "batch_size": 256, "include_mlp": false, "detach_while_caching": true, "scheduler_val_metric": ["val/accuracy", "val/IIA", "val/strict_accuracy"], "siit_sampling": "sample_all", "val_iia_sampling": "all"}

60/edges.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a2cb02e0518a50c2b53be59ea3c3788ec80c0d3c6c1e70070bd7115abdfae77
+size 113

60/ll_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddb4db579c13e1bf510ead6a01c343582eedff2f1d799f4d733d72f5f9da50ab
+size 14698

60/ll_model_cfg.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db76917a80935813594508b1d79b7a0836a20daf67e263161432430b095fd111
+size 1093

60/meta.json

@@ -0,0 +1 @@
+{"output_dir": "/circuits-benchmark/results", "atol": 0.05, "lr": 0.001, "use_single_loss": true, "iit_weight": 1.0, "behavior_weight": 0.4, "strict_weight": 0.7, "epochs": 1000, "early_stop_accuracy_threshold": 99.9, "act_fn": "gelu", "use_wandb": true, "wandb_project": "iit-train-seed-67-siit-weigth-0.7", "wandb_name": "case-60-seed-67-s-0.7-b-0.4-iit-1", "save_model_to_wandb": true, "device": "cuda", "clip_grad_norm": 0.1, "lr_scheduler": "linear", "model_pair": "strict", "same_size": false, "seed": 67, "batch_size": 256, "include_mlp": false, "detach_while_caching": true, "scheduler_val_metric": ["val/accuracy", "val/IIA", "val/strict_accuracy"], "siit_sampling": "sample_all", "val_iia_sampling": "all"}

62/edges.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a2cb02e0518a50c2b53be59ea3c3788ec80c0d3c6c1e70070bd7115abdfae77
+size 113

62/ll_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:44b75705cc0241f73d43d60bd6dba9014f5643c80a0da7d58b49ea7bdc1526fe
+size 14698

62/ll_model_cfg.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a2d11862991a5791d7bd91005f7c0f928d522a579988c7000192e53ff05de81
+size 1093

62/meta.json

@@ -0,0 +1 @@
+{"output_dir": "/circuits-benchmark/results", "atol": 0.05, "lr": 0.001, "use_single_loss": true, "iit_weight": 1.0, "behavior_weight": 0.4, "strict_weight": 0.7, "epochs": 1000, "early_stop_accuracy_threshold": 99.9, "act_fn": "gelu", "use_wandb": true, "wandb_project": "iit-train-seed-67-siit-weigth-0.7", "wandb_name": "case-62-seed-67-s-0.7-b-0.4-iit-1", "save_model_to_wandb": true, "device": "cuda", "clip_grad_norm": 0.1, "lr_scheduler": "linear", "model_pair": "strict", "same_size": false, "seed": 67, "batch_size": 256, "include_mlp": false, "detach_while_caching": true, "scheduler_val_metric": ["val/accuracy", "val/IIA", "val/strict_accuracy"], "siit_sampling": "sample_all", "val_iia_sampling": "all"}

64/edges.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a2cb02e0518a50c2b53be59ea3c3788ec80c0d3c6c1e70070bd7115abdfae77
+size 113

64/ll_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddb4db579c13e1bf510ead6a01c343582eedff2f1d799f4d733d72f5f9da50ab
+size 14698