diff --git a/.gitattributes b/.gitattributes
index 3502871b8b222f94a646a889b0eac1181c77245d..7a14fbf4206deeeb8bfb06e005e20d009cb2dd0f 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -83,3 +83,10 @@ pilecc_cont2-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge5-temperature-1.0/crit
 pilecc_cont2-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge5-temperature-1.0/actor/global_step_300/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 mathmedium2-grpo-qwen3-1.7b-4k-step-50/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_300/actor/huggingface/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/huggingface/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/huggingface/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/actor/huggingface/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/actor/global_step_100/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/actor/global_step_200/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/critic/global_step_300/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/actor/global_step_300/tokenizer.json filter=lfs diff=lfs merge=lfs -text
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_0.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_0.pt
new file mode 100644
index 0000000000000000000000000000000000000000..715335847aa5c4f8dbea29d1321922f017145eaf
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_0.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:893095e6d4a5001f70c88046eac6cd1f25221bccd19c4422deb8f2fb9dba96a1
+size 14632
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_1.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_1.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3396b28cbfc3da5d1f6ade119ca488a1eed52950
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_1.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5efafa47813652b9099b1fd2d3571809bc8d3ab2ff5b23b448595ee8b17344d4
+size 14696
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_2.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_2.pt
new file mode 100644
index 0000000000000000000000000000000000000000..ba342b2848cb152de8f656e6834c7286fb6acb65
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_2.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b711b17154f6bac80f848f97985d2dbe2afe5d93aeb4cdef185a1f4b1058f95d
+size 14696
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_3.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_3.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8439294ef0766f5862863c2031197763bf22df2c
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_3.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a94f01ceb2bc99e018478e2d79fbbcdf5c277acd6a06f8179c172193fd5083f
+size 14632
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_4.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_4.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c78b097f8819bcac8edcfb58b32799c0e451cb86
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_4.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b993a8481fe37af82869187091e7e31775ea743a5cf923d2d77c8d62a6683c6c
+size 14632
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_5.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_5.pt
new file mode 100644
index 0000000000000000000000000000000000000000..ba1ea08fa6a73d67603f4b57a3b8eceb2f6c9a67
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_5.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:829ce983a89f06a5db53a389ccff64f2eca40f5a035f1d4acacc28d06b8d8f3e
+size 14696
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_6.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_6.pt
new file mode 100644
index 0000000000000000000000000000000000000000..35924ea6c6b50f4b004b78af833d0693b7bfde88
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_6.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aba6d27a58bbe25fdff9eab7483fc4470fe462e85b33e3016754adec2de933f8
+size 14696
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_7.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_7.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d0d701c45a545d72c96b06e71c7e75ccaf012407
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/extra_state_world_size_8_rank_7.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e2b37dcd197e2605c4bb32a59e9672345cc323097402eedc0fb0c4988164bac
+size 14632
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/huggingface/model-00001-of-00003.safetensors b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/huggingface/model-00001-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..c580e81c5da78c303b011666a9799dfe73139902
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/huggingface/model-00001-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:974eca4df05a71295b229fd1896f66121939cd00f912a505f747964f53290074
+size 4998767360
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/huggingface/model-00002-of-00003.safetensors b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/huggingface/model-00002-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..ab37571b9164d646043eff0a030519325ec7d4fc
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/huggingface/model-00002-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9cd2ae5820c2c4b76df65550b2cd1bbc2ee4c59f379252db1250691384ecae4d
+size 4932808968
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/huggingface/model-00003-of-00003.safetensors b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/huggingface/model-00003-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..e8cef48a92a96725cdc163b2a5ced4b707043597
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/huggingface/model-00003-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:431d5db10a6af75880102621916fdedb5babcb2cb27c553868c0dcd781843a18
+size 4495462048
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/huggingface/tokenizer.json b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/huggingface/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..1c1d8d5c9024994f1d3b00f9662b8dd89ca13cf2
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/huggingface/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b9e4e7fb171f92fd137b777cc2714bf87d11576700a1dcd7a399e7bbe39537b
+size 17209920
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_0.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_0.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6d8568f3ac1f0645d5290a06c463bc441c8edf5f
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_0.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0803c511790e0cbf463dc8500b446c4170847a05e9f306a88d8b103f70ecd611
+size 1803528186
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_1.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_1.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1c793545a306bee8c0fcd93fc97638f008e67635
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_1.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:882a5f934b890fda52663dcd230fa8774134cebcffb8f16482c664117e1339d2
+size 1803528186
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_2.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_2.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8c4ee04971b602be0246ecfab795235b39f489aa
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_2.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:227893ba66fcb5c2df6f8a17ea2a08e279e8a26f3cffaafe72266a183be4141b
+size 1803528186
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_3.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_3.pt
new file mode 100644
index 0000000000000000000000000000000000000000..446943f2072e476a80afd9b51e3ff8ba4de192f4
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_3.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbae17dc4830ebf3e0ecb55da51baf5dbe56f222a5b8ba7fe242d2431b58091c
+size 1803528186
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_4.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_4.pt
new file mode 100644
index 0000000000000000000000000000000000000000..de7e19b6b9063dc3a33ae4740d0f6d0ccf2e66ee
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_4.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33781bff0f3301eb8bcb0dbdc4742f35ebcc5840ea0cf4ff4cad6b0c2e81bc0b
+size 1803528186
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_5.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_5.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6cfce577a8a62b377757e0e709e37a81d5aaa0c7
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_5.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e20b8133290d5f0b9e4ec70e5b3ed5c09a917a307b83e7b7748412e21c32185
+size 1803528186
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_6.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_6.pt
new file mode 100644
index 0000000000000000000000000000000000000000..29fc3079700ef3f59924031af702e868c90b1da6
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_6.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da114725dc6086aa5c7b9bf08502666b5043860b12301e59a2a19cd696f3bd66
+size 1803528186
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_7.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_7.pt
new file mode 100644
index 0000000000000000000000000000000000000000..633a2248daa7b94746cc3fcd9c22f314437edac0
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/actor/model_world_size_8_rank_7.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39b6bfa49c21e241d2cd4368180d90167c3c9e358924de1bd9c96dede6be18e2
+size 1803528186
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/data.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/data.pt
new file mode 100644
index 0000000000000000000000000000000000000000..22064a19b8f2079f90a34f1c2cff534b1abbf2f5
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_100/data.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3316b4d2ad4febf3149e432bdb73072d8cfde197f339611644643ae5b0ce119c
+size 1492
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/actor/huggingface/model-00001-of-00003.safetensors b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/actor/huggingface/model-00001-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..fdaae08b3b018f53c6e8cb0a3e3630e159f69b18
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/actor/huggingface/model-00001-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0ab878d379f9387e00d9f34fbf29de05c19b97e932c81790a3a37c90379a6e98
+size 4998767360
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/actor/huggingface/model-00002-of-00003.safetensors b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/actor/huggingface/model-00002-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..394ef6b7304011dfccc5f81e5050cb152dec0ee1
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/actor/huggingface/model-00002-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ab999440a18951ce5667f256321264bd84e2092acc0677683a8c4023e0a7e3c
+size 4932808968
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/actor/huggingface/model-00003-of-00003.safetensors b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/actor/huggingface/model-00003-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..189fac5ad5c178b5bcf902cadaba52b7177738a7
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/actor/huggingface/model-00003-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd16704e39fcacd2036faa82fcf068286cb4b1ad323bb407cd3f3ed859a0cc5a
+size 4495462048
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/actor/huggingface/tokenizer.json b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/actor/huggingface/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..1c1d8d5c9024994f1d3b00f9662b8dd89ca13cf2
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/actor/huggingface/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b9e4e7fb171f92fd137b777cc2714bf87d11576700a1dcd7a399e7bbe39537b
+size 17209920
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/data.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/data.pt
new file mode 100644
index 0000000000000000000000000000000000000000..ce22321628ed55ad7970ab89a9fa9834a64402be
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_200/data.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2d9905b441d82a48e798180d0463e7858c0abf50c33e5e83e44da9135bb0060
+size 1492
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/extra_state_world_size_8_rank_2.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/extra_state_world_size_8_rank_2.pt
new file mode 100644
index 0000000000000000000000000000000000000000..42c68f37fe35d94916653939b5b4c9d25e311b3c
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/extra_state_world_size_8_rank_2.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:96a6bc3a214283dae13157462afaa5b4b2326cbcd526bbe094b4818ecd5e8eee
+size 14632
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/extra_state_world_size_8_rank_5.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/extra_state_world_size_8_rank_5.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2e4a38292c9fb3c1ae26b3f0e9bae97bcf6aff41
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/extra_state_world_size_8_rank_5.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:926a772a7e2249e78a2959eaa74ee1db4cca0535ec43b9a5fc06fa55f13e7870
+size 14696
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/huggingface/model-00001-of-00003.safetensors b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/huggingface/model-00001-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..3c1924e10f9b4c051cf1b119afe2c604fbf07bef
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/huggingface/model-00001-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee2a1f8f494de18824621f0b5a8e493fc527c9323ba65ae49c70beed65e9f564
+size 4998767360
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/huggingface/model-00002-of-00003.safetensors b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/huggingface/model-00002-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..3a1cf8989f17b962e5ec7dd85932cee63432778b
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/huggingface/model-00002-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2a1462403ad83bb529f72e5df24929436b34a6af83c4d070e139ba94761aef8
+size 4932808968
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/huggingface/model-00003-of-00003.safetensors b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/huggingface/model-00003-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..2d236c537ab0ffcd4ca37d1bc5fc8ec98a467ff8
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/huggingface/model-00003-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72e6fe81a3a12cc44398c9f7cd492f0a867252b068ea8dd1ddc624caa47689e6
+size 4495462048
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/huggingface/tokenizer.json b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/huggingface/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..1c1d8d5c9024994f1d3b00f9662b8dd89ca13cf2
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/huggingface/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b9e4e7fb171f92fd137b777cc2714bf87d11576700a1dcd7a399e7bbe39537b
+size 17209920
diff --git a/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/model_world_size_8_rank_0.pt b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/model_world_size_8_rank_0.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4b07a87177657154d19286f2643c78758b4f07fe
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-llama3.2-3b-it-oldreward-4k-corrupt_delete-0.2/global_step_250/actor/model_world_size_8_rank_0.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e02155fec3aa6a5c11a54d803e35a78c22504ddbce88fba26b549f6612d13c8
+size 1803528186
diff --git a/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/actor/global_step_100/model-00001-of-00003.safetensors b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/actor/global_step_100/model-00001-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..8dc95510ef38097f937b502b0653d8f62540127b
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/actor/global_step_100/model-00001-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f13c6f5d4ce89d64c27b23539266705299107e053beea4fdda0854d76951731
+size 4982131536
diff --git a/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/actor/global_step_100/model-00002-of-00003.safetensors b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/actor/global_step_100/model-00002-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..16d3e79a41a3b329f76834733813e7ae4329218d
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/actor/global_step_100/model-00002-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f27c47764e0d894ab9c22e161db0e72a20b1d07c7a41568a5eecb11ccf3f835e
+size 4932949336
diff --git a/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/actor/global_step_100/model-00003-of-00003.safetensors b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/actor/global_step_100/model-00003-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..95cafbfc19d88ba0a0bc7fc2a82513834ca60e30
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/actor/global_step_100/model-00003-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ddff00d062d4ffd517d5b25aeb996bc5ebb9b9c3d038568b29315df65e2824a
+size 3673383040
diff --git a/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/actor/global_step_100/tokenizer.json b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/actor/global_step_100/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..51ebb3ba93988c73bbd83a3c163a14343fbc5b06
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/actor/global_step_100/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
+size 11421896
diff --git a/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/random_state.pkl b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/random_state.pkl
new file mode 100644
index 0000000000000000000000000000000000000000..66bdfdd8c28ec19192ef0a73d627d1c252b9752c
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_100/random_state.pkl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fc55ffe1e131cf4556078de9e5d14e45864dd9f0b0c3e2439967761a98dcb1d0
+size 12846
diff --git a/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/actor/global_step_200/model-00001-of-00003.safetensors b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/actor/global_step_200/model-00001-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..7e98d372702149f55c755f94b4ae05b6cd40d9d4
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/actor/global_step_200/model-00001-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b52763d1189b28c46c34b39bed5efdc44c12ddb9ddf432a3552046ba95067c9
+size 4982131536
diff --git a/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/actor/global_step_200/model-00002-of-00003.safetensors b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/actor/global_step_200/model-00002-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..6be3f2124723f1c4a337f5ede1632095437ad6c2
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/actor/global_step_200/model-00002-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4dd978e275f23a3faeb993cbfc66b7ad5b0433294688c3ded53030535ed9eec
+size 4932949336
diff --git a/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/actor/global_step_200/model-00003-of-00003.safetensors b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/actor/global_step_200/model-00003-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..38cc35d01de37296dc1652e0f6c4c73dcf44d4a6
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/actor/global_step_200/model-00003-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ea006bdb65c2c3feecf876b15621bb1547981b0b1f5a146246c983df7726d3cc
+size 3673383040
diff --git a/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/actor/global_step_200/tokenizer.json b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/actor/global_step_200/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..51ebb3ba93988c73bbd83a3c163a14343fbc5b06
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/actor/global_step_200/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
+size 11421896
diff --git a/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/random_state.pkl b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/random_state.pkl
new file mode 100644
index 0000000000000000000000000000000000000000..39aa837fa2c415e35eda6b3f80b589b7383c5b43
--- /dev/null
+++ b/mathmedium2-mutualposclip0.5-qwen2.5-3b-oldreward-corrupt-0.1/global_step_200/random_state.pkl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:066b8bf0e896ba7739b3639a14e770552dca52904c02795ee8bca574c21e3180
+size 12846
diff --git a/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/actor/global_step_300/model-00001-of-00003.safetensors b/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/actor/global_step_300/model-00001-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..5b7169ebff851eb52acb4d3b4a0ad834786c8eec
--- /dev/null
+++ b/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/actor/global_step_300/model-00001-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4989f3161f61135573d1d777a424a7e290644b05f5b3d76376b2b877cf90f89
+size 4998767360
diff --git a/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/actor/global_step_300/model-00002-of-00003.safetensors b/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/actor/global_step_300/model-00002-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..f5e640db1018b5cb5f8beeda201421b9bc4d3232
--- /dev/null
+++ b/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/actor/global_step_300/model-00002-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54dcb74c1756b74f4c23a79a250806395722e59c2b8b458067537c5984580fb4
+size 4932808968
diff --git a/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/actor/global_step_300/tokenizer.json b/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/actor/global_step_300/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..1c1d8d5c9024994f1d3b00f9662b8dd89ca13cf2
--- /dev/null
+++ b/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/actor/global_step_300/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b9e4e7fb171f92fd137b777cc2714bf87d11576700a1dcd7a399e7bbe39537b
+size 17209920
diff --git a/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/critic/global_step_300/model-00001-of-00003.safetensors b/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/critic/global_step_300/model-00001-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..9d1ef462bb5fb4946501111447c394bc1d30d083
--- /dev/null
+++ b/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/critic/global_step_300/model-00001-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:426e03b27cbb4a052599448df1f2ebbe1859260818706cee0f0cebdd3eabf6fe
+size 4998767360
diff --git a/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/critic/global_step_300/model-00002-of-00003.safetensors b/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/critic/global_step_300/model-00002-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..23d8945ed0170313560fdf36b3fe2999a8c296d7
--- /dev/null
+++ b/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/critic/global_step_300/model-00002-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c08c6a1702ad1c31f562fd5e5fb900c61ea03c0690c9f5793ea505ac4a128252
+size 4932808968
diff --git a/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/critic/global_step_300/model-00003-of-00003.safetensors b/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/critic/global_step_300/model-00003-of-00003.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..50d329eec4d7e8a0a46e539228fdc7677aa7deab
--- /dev/null
+++ b/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/critic/global_step_300/model-00003-of-00003.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b7b6abccf30397a64a89fb0e5bd7b6dc1e7bd14589276fbb7bc855b2073b45de
+size 2919464628
diff --git a/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/critic/global_step_300/tokenizer.json b/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/critic/global_step_300/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..1c1d8d5c9024994f1d3b00f9662b8dd89ca13cf2
--- /dev/null
+++ b/pilecc_cont1-r1-ppo-llama3.2-3b-em-warmup-0.05-rouge-rouge3-temperature-1.0/critic/global_step_300/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b9e4e7fb171f92fd137b777cc2714bf87d11576700a1dcd7a399e7bbe39537b
+size 17209920