diff --git a/.gitattributes b/.gitattributes
index 43e3f566d92e04c595e1e1cd1340b578761e042f..677c8c8bf26e9c10a12fc7f9e6920ac87f7c8361 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -959,3 +959,38 @@ cache_nq_qwen4b/20260504_160206/rollout/226.jsonl filter=lfs diff=lfs merge=lfs
 cache_nq_qwen4b/20260504_160206/rollout/230.jsonl filter=lfs diff=lfs merge=lfs -text
 cache_nq_qwen4b/20260504_160206/rollout/242.jsonl filter=lfs diff=lfs merge=lfs -text
 cache_nq_qwen4b/20260504_160206/global_step_250/actor/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/232.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/223.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/246.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/245.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/220.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/216.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/117.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/248.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/120.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/107.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/233.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/252.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/116.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/105.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/109.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/108.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/115.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/121.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/103.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/122.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/222.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/237.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/243.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/119.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/224.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/231.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/234.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/235.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/247.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/217.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/240.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/239.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/225.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/251.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_nq_qwen4b/20260504_160206/rollout/214.jsonl filter=lfs diff=lfs merge=lfs -text
diff --git a/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/103.jsonl b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/103.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..aa0e187ba5ee9388d13bf69537ae12056a41cde0
--- /dev/null
+++ b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/103.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2352572f6dd229fcd5e9353479378644d1e4c943a727fc0736a8f85280bb4bc9
+size 14546528
diff --git a/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/105.jsonl b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/105.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..9ffdea2176b001ee2a99e72365763e04067d0192
--- /dev/null
+++ b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/105.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:152e4826a52cf438893fe250e0cee35260a7d132d9a69b450a8b9332303ee936
+size 14055491
diff --git a/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/107.jsonl b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/107.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..9fb13a016e5039517c906a2f7fe060266a93455a
--- /dev/null
+++ b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/107.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c932cce7ea88d8bdce9196d434f07724db196212d38963c665de2075436ec02
+size 15082902
diff --git a/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/108.jsonl b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/108.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..40b7dc0f1ed5aada02af9086dd33905f773d6846
--- /dev/null
+++ b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/108.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a46fd124244fff4e8bffc74bc0aec2090d4ae97c65c3a21876e11319f9a3d5e
+size 14942217
diff --git a/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/109.jsonl b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/109.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..5a0501e2dd3013bea74db0532cfa39c80eba7c19
--- /dev/null
+++ b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/109.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:495e2abde010b06190031dd9553f73e043169b183567b547fa9f9d361ab9ff7f
+size 14621388
diff --git a/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/115.jsonl b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/115.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..5b57b2176c5058c1ef97a97478e9d6b8e5994b80
--- /dev/null
+++ b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/115.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c3b6a8dd8ec41136a1ddfeece20c245a6c0853a85d7c1eccea95f3994ab1a433
+size 13449926
diff --git a/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/116.jsonl b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/116.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..4ba39dd55f71823908fc9d4c1e3e53bbd5a865fc
--- /dev/null
+++ b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/116.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de2fe06a9c7fbf8e848b2dde7f1eca0245238a81b5f4ac02968df07092746696
+size 12383918
diff --git a/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/117.jsonl b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/117.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..67e61e88d35f3fe68ddd8c2fcab543c31bb5fbe0
--- /dev/null
+++ b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/117.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a36633f480f8c55af9a5d81ee11703f1855dcdd5541d21e50d59455528e12aa9
+size 13287264
diff --git a/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/119.jsonl b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/119.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..d7c40e1352b4b631ced1a0ee8619516f561f9905
--- /dev/null
+++ b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/119.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:94057c585875742cb11816458cd9bfc9f9598625c5f3427eda7b33b329fdaf0e
+size 14771823
diff --git a/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/120.jsonl b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/120.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..f91c87b3336bbc3ce080de5e8704a0b6ec4a71c6
--- /dev/null
+++ b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/120.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0eb2b3929585ee71471497041e01c5c4087764e2871e1541d5c4c4e1c7690911
+size 14724128
diff --git a/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/121.jsonl b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/121.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..15905467dd35e299a29d6ece793363f25a1d47d6
--- /dev/null
+++ b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/121.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ebeee8a150c4b21f1ee9768dd6704ae6ae6dba893954c6e67295c70852fa5948
+size 14115448
diff --git a/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/122.jsonl b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/122.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..b0b58b26627b1ecb49223c2a67a7584322c917f5
--- /dev/null
+++ b/cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/122.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0957697eb6cc2a6a53513885afc38531573bdd28121a3c86e6cf7a415c1aca1
+size 13743775
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/extra_state_world_size_8_rank_1.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/extra_state_world_size_8_rank_1.pt
new file mode 100644
index 0000000000000000000000000000000000000000..63ba3894a4e8bc277e38b8663351c8da9e6e48dd
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/extra_state_world_size_8_rank_1.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed16a8eeb716045b1d5e340e5b092cb348c4130394c58f5028089ecf4ce98e1f
+size 14632
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/extra_state_world_size_8_rank_5.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/extra_state_world_size_8_rank_5.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5c2239114dd337fc11a620f9da54d17ec72c733f
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/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 sha256:9555d4f192660933e164807d87641992b0d03cf16ee5c29ee322a96332d100a3
+size 14632
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_0.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_0.pt
new file mode 100644
index 0000000000000000000000000000000000000000..25c0e842618d3a2fe89dbbdfe2a29a0563766169
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_0.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e973efab8a80ad6d2626ec68319ecfd79fe3ee872531338b4431805311b3ffd
+size 2205945402
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_1.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_1.pt
new file mode 100644
index 0000000000000000000000000000000000000000..97304a9fbe2b057410fda9ae1718613b850e1ce0
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_1.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:20c636ec16345bd609e0450edde06814a2f51f095e644093fe1235d45babc0f1
+size 2205945402
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_2.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_2.pt
new file mode 100644
index 0000000000000000000000000000000000000000..284eaf98be8cc61f70b026f34cc08aca78f15caf
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_2.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:336bdea46a6505c14cfa7292afe3bf591b9017314435dd66c5d5c5d80b2c8c35
+size 2205945402
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_3.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_3.pt
new file mode 100644
index 0000000000000000000000000000000000000000..63a88126d40063963148a052b8265f42dc78e00e
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_3.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a9c64d369a1832009d6270fb2adc3d32ad997670a6e9722d69a0a11521febc7
+size 2205945402
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_4.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_4.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1b1db902fc5e1f13d25b049c27c3844e6b4e8930
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_4.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e42b89b57afc547df1b1d6b70ad19d5dc71495ca049c3e9c07b8defbe804a218
+size 2205945402
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_5.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_5.pt
new file mode 100644
index 0000000000000000000000000000000000000000..cc49c1cd9f9989f1299bf8757ed3672a70215798
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_5.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:750744f724163380d5a0907ccb5c7231c328a8d3236b946448888bc1e0672a6b
+size 2205945402
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_6.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_6.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7dda54c051958284dfad2c82eaf8b174d0ffd85e
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_6.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29daf20d8901aca8a5d6c5eb30a7990877b981328bdb52399c7bfb52e09af937
+size 2205945402
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_7.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_7.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6dde14d7ff2f39f28c557d9c18f284292f9ed84b
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/model_world_size_8_rank_7.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6dfbcf955a369e880859afac65eb9004611de21fe2ab0bd104c6312575dd38d2
+size 2205945402
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_0.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_0.pt
new file mode 100644
index 0000000000000000000000000000000000000000..665316cb10280226af6e8c2da5f0c13f224cfedb
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_0.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c642fce90410a736273199f92335e42a3d7c4a172de8bf911d17c9f6b61a1e9
+size 4022501290
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_1.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_1.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1e30ec61be82b0fc3d6720318e1064bcceb52c08
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_1.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0f2ba066313dcdb64a9913450ed80c9e1dd0a0be3dbe4d483a04c2e0aa932c58
+size 4022501290
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_2.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_2.pt
new file mode 100644
index 0000000000000000000000000000000000000000..9627e8e737e3dbf5158a87a929d9cb80084bf161
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_2.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08a05deed1768c7f4a67d149cc8384451c2b19359d38493b6c546679791115b4
+size 4022501290
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_3.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_3.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a31b966155bf8030ba0755de348417b431e6728a
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_3.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9007b5853d4ffca17e0224725d7a7fadcf01c7c47c5be8e09566551c0e862efc
+size 4022501290
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_4.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_4.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7e1ab8ed1dc34211cad5130cdc5c3d25f6e5a3a4
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_4.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cbb507b3d9e7943e03e9c86b59be02175d3f09d70075122320128bea951a7469
+size 4022501290
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_5.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_5.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b6c6a8de5ce15050d6368a45b932aab31e3f0744
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_5.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:786ff3ff634ada2998502ec7016a148793c80b34cbafeb410917f02dab0c65d5
+size 4022501290
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_6.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_6.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8a12c78a83f130ac7aac8690066164d52adb72d9
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_6.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b7deac67b983c6ad2c83f3fd48222118b93c511681648894708b56ec1c029b3
+size 4022501290
diff --git a/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_7.pt b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_7.pt
new file mode 100644
index 0000000000000000000000000000000000000000..bf3656a9701644aa61eca784e9ca0dd9a1b20840
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/global_step_250/actor/optim_world_size_8_rank_7.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9fe421e18ac41c0a83af033b215a1d624cdb9d808034533e1baa8ccd50d3fc22
+size 4022501290
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/214.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/214.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..09eabbc589a7c6a5894cef3a2a38373be1b36f96
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/214.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e15dde075cc3ae2c14fa8af5c0947bb167b4837de24eef75af844d0f959f562
+size 12189240
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/216.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/216.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..7ce62223c69a1d56520671b8b636948bb5a09160
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/216.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cbaca8d1e316fda7bed2f951bd24e6fe377db617fe8bf6abc5ae3d49dd557bd0
+size 11648569
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/217.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/217.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..1f4f8c33e7c421e94527f6a112cc2dd0638af835
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/217.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:261212eb3184607168aea15887172903389b86cf44bfa640eebcf894a8bfae1d
+size 12629252
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/220.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/220.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..d2f315d6db6baf2341b3ddb25d7671c3010da5f0
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/220.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:318f847ac834b9e937b3f589444a3e9cbb3555e08a92467380d2a27d722be07f
+size 11887456
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/222.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/222.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..b282fb38234f1317b95c776b43f55215e1105b33
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/222.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de77b2911daac7c0e1f7703f5dabacb3cff83664697606ff2e9684dcae644034
+size 12243149
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/223.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/223.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..eb131cd1628bd51ae1e747bb18ed11a7f5c7e1a5
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/223.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d9ee6ad0bb5a7fc7bac4348e0f4cabd823b1ada9aae0b077ee02adf2d50dbeee
+size 12571251
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/224.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/224.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..fb64574982179fb5d9f56d4e60a504e5d2129f72
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/224.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06dce0785104b287922cd08b91cfd31f07b3104c0d0ec803491aaddc4a03613e
+size 12493340
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/225.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/225.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..ae7780473651d02e650afb0b0e3d8a09ea6ca72c
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/225.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c6efe52e0e0f4da70e7763d1f2ca141b477c576b02f85e8974724327c4ff3e8e
+size 13139263
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/231.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/231.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..b6bb9763967a122067ae40291fd5b7da270397c0
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/231.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37bb0c74a1f465bef66597c3090fc161e51cfcbff9ce766666da59626039fbf3
+size 12415455
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/232.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/232.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..e796327b394877a3719d384abe3b594952543c92
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/232.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e074da13ad4e3ea668f502e3579deb4787f3bce5202ecca4d551a09f01e8b2b0
+size 11834130
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/233.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/233.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..951fceb363a06e14260b41556eee6500aa2db8e1
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/233.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ba5aa5330695e36e2422050b247cba5ba09fb67d249f883a1ec7f9eee59777b
+size 11831530
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/234.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/234.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..fec20213c1bf4aeb9338680aa7d3b30ab4270757
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/234.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:576014bcff655f49b420a3ea09a9b612c053167884d3cf2d5819d4ad22e5ecf0
+size 11000288
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/235.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/235.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..140ef4508a12cedf098a0e22b610063136a97729
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/235.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:542005ef7bd00f71d8e48181d6c92f84e23c8700fdac2a11e5e095acc70046b5
+size 10800056
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/237.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/237.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..be0ee2bf5d3143b61973ecbda3e387efb1f86642
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/237.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35c452ee9bd9d6681415297058ddd327cfb47848ecafefd974c8ef46a2eb5df0
+size 10974873
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/239.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/239.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..123454bf030e7e0e9c1dc081d2eeb8b0d60dcea5
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/239.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b59b554701642f9c4bff71b411ed77931f2bca422aa27244d945f5622e84d227
+size 11318530
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/240.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/240.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..c750b99dd4f9b5a7ebe451097772e4fd1a3dd64b
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/240.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a08c80a749547ec503d0aca4e8ab1ada8d18afa2f1f3ffede6846318ccaac64c
+size 12186786
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/243.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/243.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..4707d122c84024e8bd7a2a526c1024ae6aa4406b
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/243.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba5cf7acc2627a82ca7ec790ec9b5627d98fb6c0ba69349c78ef47633bac358e
+size 12021451
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/245.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/245.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..bdc2a334a5e0be4cc30f366cbafabcb444dd9297
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/245.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4438060aed2af50c436be0c091f7395e5f60e0dc3c169e6f8e0231d83e4098a8
+size 11512306
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/246.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/246.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..ad56164ca035c71f2065a03eb4c63005718b11af
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/246.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52d84c83cd6a8d9c3d832f9d84b71131c2112bc445c9c2fc90f0e20df8c83c0e
+size 11125116
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/247.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/247.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..6cedec633ad0c5a021759bc7b42f2a6630f00eca
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/247.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0a0b52b2a3642dbd8fa6b4db2cc6ad565b49aae3fa340306567ba09decf2eec
+size 11940566
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/248.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/248.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..ddea878fd7737483917c9d22f5b3de5c344ba07c
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/248.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1cd3b437b3c9435bd0f14d42a8be5f248308d11337554cd7d02713788daa5220
+size 12042845
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/251.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/251.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..ca595b44306a5c4b63f76ef9d62bfc528a5c1d7c
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/251.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89bde6bcdca2609c57e0784619a1567341aa8168a5a7f154dd25ad9d65f9e327
+size 12434047
diff --git a/cache_nq_qwen4b/20260504_160206/rollout/252.jsonl b/cache_nq_qwen4b/20260504_160206/rollout/252.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..b0c92c6a3c90e736944fe36e5168c43c82a43bb3
--- /dev/null
+++ b/cache_nq_qwen4b/20260504_160206/rollout/252.jsonl
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ea9b12c2247308af54f67a64f322bf45feb5e331e1a6a0e26b51088e4f1915c
+size 11892889