Add files using upload-large-folder tool

.gitattributes

@@ -994,3 +994,33 @@ cache_nq_qwen4b/20260504_160206/rollout/239.jsonl filter=lfs diff=lfs merge=lfs
 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
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/209.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/129.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/157.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/208.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/237.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/221.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/143.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/213.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/160.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/195.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/215.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/177.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/178.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/185.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/156.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/176.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/130.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/163.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/150.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/200.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/199.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/194.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/233.jsonl filter=lfs diff=lfs merge=lfs -text
+cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/136.jsonl filter=lfs diff=lfs merge=lfs -text
+atpo_hotpotqa_qwen4b/20260506_102808/rollout/235.jsonl filter=lfs diff=lfs merge=lfs -text
+atpo_hotpotqa_qwen4b/20260506_102808/rollout/232.jsonl filter=lfs diff=lfs merge=lfs -text
+atpo_hotpotqa_qwen4b/20260506_102808/rollout/231.jsonl filter=lfs diff=lfs merge=lfs -text
+atpo_hotpotqa_qwen4b/20260506_102808/rollout/240.jsonl filter=lfs diff=lfs merge=lfs -text
+atpo_hotpotqa_qwen4b/20260506_102808/rollout/222.jsonl filter=lfs diff=lfs merge=lfs -text
+atpo_hotpotqa_qwen4b/20260506_102808/rollout/220.jsonl filter=lfs diff=lfs merge=lfs -text

atpo_hotpotqa_qwen4b/20260506_102808/global_step_250/actor/optim_world_size_8_rank_0.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7573736c838c607348eec7d642227db2a8720a5235f429c8912dc52ba8d37371
+size 4022501290

atpo_hotpotqa_qwen4b/20260506_102808/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:c716c3cdf6ad8f7b3e86e695f015c681cf42f60200c98a8bf12204dd85d0270f
+size 4022501290

atpo_hotpotqa_qwen4b/20260506_102808/global_step_250/actor/optim_world_size_8_rank_5.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b3dde2b35db02b1206aae69aa4225e46aec6a68c07705ca6df36303102d1c87
+size 4022501290

atpo_hotpotqa_qwen4b/20260506_102808/outputs/.hydra/overrides.yaml

@@ -0,0 +1,70 @@
+- algorithm.adv_estimator=grpo
+- algorithm.kl_ctrl.kl_coef=0.0
+- data.train_files=/scratch/user/sushil22_tamu.edu/projects/ATPO/rl_datasets/hotpotqa/train.parquet
+- data.val_files=["/scratch/user/sushil22_tamu.edu/projects/ATPO/rl_datasets/hotpotqa/test_512.parquet"]
+- data.prompt_key=prompt
+- data.train_batch_size=64
+- data.max_prompt_length=2000
+- data.max_response_length=6192
+- actor_rollout_ref.model.path=/scratch/user/sushil22_tamu.edu/projects/ATPO/models/Qwen3-4B
+- actor_rollout_ref.model.enable_gradient_checkpointing=True
+- actor_rollout_ref.model.use_remove_padding=True
+- actor_rollout_ref.actor.policy_loss=gspo_turn
+- actor_rollout_ref.actor.clip_ratio_low=3e-3
+- actor_rollout_ref.actor.clip_ratio_high=4e-3
+- actor_rollout_ref.actor.enable_entropy_balanced_clipping=False
+- actor_rollout_ref.actor.enable_entropy_balanced_advantage=False
+- actor_rollout_ref.actor.optim.lr=1e-6
+- actor_rollout_ref.actor.ppo_mini_batch_size=8
+- actor_rollout_ref.actor.use_dynamic_bsz=True
+- actor_rollout_ref.actor.ppo_max_token_len_per_gpu=16384
+- actor_rollout_ref.actor.use_kl_loss=True
+- actor_rollout_ref.actor.kl_loss_coef=0.0
+- actor_rollout_ref.actor.kl_loss_type=low_var_kl
+- actor_rollout_ref.actor.fsdp_config.param_offload=False
+- actor_rollout_ref.actor.fsdp_config.optimizer_offload=False
+- actor_rollout_ref.rollout.enable_dynamic_rollouts=False
+- actor_rollout_ref.rollout.log_prob_max_token_len_per_gpu=32768
+- actor_rollout_ref.rollout.tensor_model_parallel_size=1
+- actor_rollout_ref.rollout.name=vllm
+- actor_rollout_ref.rollout.mode=sync_with_tool_tree
+- actor_rollout_ref.rollout.gpu_memory_utilization=0.6
+- actor_rollout_ref.rollout.n=22
+- actor_rollout_ref.rollout.initial_rollouts=10
+- actor_rollout_ref.rollout.expansion_mode=entropy
+- actor_rollout_ref.rollout.expansion_iterations=2
+- actor_rollout_ref.rollout.beam_size=6
+- actor_rollout_ref.rollout.samples_per_tree=22
+- actor_rollout_ref.rollout.branch_probability=0.5
+- actor_rollout_ref.rollout.entropy_weight=0.2
+- actor_rollout_ref.rollout.leaf_value_norm=True
+- actor_rollout_ref.rollout.node_value_mode=child_softmax
+- actor_rollout_ref.rollout.node_adv_mode=node_value
+- actor_rollout_ref.rollout.tools.call_limit=6
+- ++actor_rollout_ref.rollout.tools.tool_instances.search.params.cache_file=/scratch/user/sushil22_tamu.edu/projects/ATPO/search_cache/search_cache.json
+- ++actor_rollout_ref.rollout.tools.tool_instances.search.params.api_key=unused_local_rag_server
+- actor_rollout_ref.rollout.multi_turn.enable=True
+- actor_rollout_ref.rollout.multi_turn.tool_config_path=/scratch/user/sushil22_tamu.edu/projects/ATPO/ATPO/verl_atpo/examples/sglang_multiturn/config/tool_config/search_tool_config.yaml
+- actor_rollout_ref.ref.log_prob_max_token_len_per_gpu=32768
+- actor_rollout_ref.ref.fsdp_config.param_offload=True
+- reward_model.reward_manager=naive
+- custom_reward_function.path=/scratch/user/sushil22_tamu.edu/projects/ATPO/ATPO/verl_atpo/verl/utils/reward_score/deep_research_em.py
+- custom_reward_function.name=compute_score
+- trainer.critic_warmup=0
+- trainer.logger=[console, wandb]
+- trainer.project_name=Neurips-final-runs
+- trainer.experiment_name=atpo_hotpotqa_qwen4b_ablation_noadvantage
+- trainer.n_gpus_per_node=8
+- trainer.nnodes=1
+- trainer.save_freq=50
+- trainer.test_freq=10
+- trainer.max_actor_ckpt_to_keep=2
+- +actor_rollout_ref.actor.checkpoint.save_contents=[hf_model,extra]
+- trainer.total_epochs=1
+- trainer.default_local_dir=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen4b_ablation_noadvantage/20260506_102808/
+- trainer.val_before_train=True
+- trainer.rollout_data_dir=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen4b_ablation_noadvantage/20260506_102808//rollout
+- trainer.validation_data_dir=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen4b_ablation_noadvantage/20260506_102808//validation
+- ray_init.num_cpus=null
+- trainer.resume_mode=resume_path
+- trainer.resume_from_path=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen4b/20260430_082825/global_step_200

atpo_hotpotqa_qwen4b/20260506_102808/rollout/220.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ea3cd3f6d23f18ea1858066864051a264effe053e0519610d64ef0f15b77fc9
+size 14714497

atpo_hotpotqa_qwen4b/20260506_102808/rollout/222.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88ee49045b8af63c5b4999b806019a1e48233e7092558a56b1cf72644851b926
+size 13183027

atpo_hotpotqa_qwen4b/20260506_102808/rollout/231.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02495cfd1f50cb81e1ada4a561143f7c26036c63d93c501fe60f20b1a60ada57
+size 14235762

atpo_hotpotqa_qwen4b/20260506_102808/rollout/232.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd5453481bde797d6c360a3c58571d0b9ab1f2eb355ea50cb310399404345ae2
+size 13841430

atpo_hotpotqa_qwen4b/20260506_102808/rollout/235.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a88baa8399c639b0856cd3cb99eb11548d082753cd6bf5231e737f5286c13e1d
+size 15103381

atpo_hotpotqa_qwen4b/20260506_102808/rollout/240.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02859f96249c3db635d7e112a85c9b5f252526e547df1b8370c2f00936e7c847
+size 13592804

atpo_hotpotqa_qwen8b/20260506_124901/outputs/.hydra/config.yaml

@@ -0,0 +1,319 @@
+data:
+  tokenizer: null
+  use_shm: false
+  train_files: /scratch/user/sushil22_tamu.edu/projects/ATPO/rl_datasets/hotpotqa/train.parquet
+  val_files:
+  - /scratch/user/sushil22_tamu.edu/projects/ATPO/rl_datasets/hotpotqa/test_512.parquet
+  prompt_key: prompt
+  reward_fn_key: data_source
+  max_prompt_length: 2000
+  max_response_length: 6192
+  train_batch_size: 64
+  val_batch_size: null
+  return_raw_input_ids: false
+  return_raw_chat: false
+  return_full_prompt: false
+  shuffle: true
+  filter_overlong_prompts: false
+  filter_overlong_prompts_workers: 1
+  truncation: error
+  image_key: images
+  video_key: videos
+  trust_remote_code: false
+  custom_cls:
+    path: null
+    name: null
+algorithm:
+  gamma: 1.0
+  lam: 1.0
+  adv_estimator: grpo
+  norm_adv_by_std_in_grpo: true
+  use_kl_in_reward: false
+  kl_penalty: kl
+  kl_ctrl:
+    type: fixed
+    kl_coef: 0.0
+    horizon: 10000
+    target_kl: 0.1
+  use_pf_ppo: false
+  pf_ppo:
+    reweight_method: pow
+    weight_pow: 2.0
+actor_rollout_ref:
+  hybrid_engine: true
+  model:
+    path: /scratch/user/sushil22_tamu.edu/projects/ATPO/models/Qwen3-8B
+    use_shm: false
+    external_lib: null
+    override_config: {}
+    enable_gradient_checkpointing: true
+    enable_activation_offload: false
+    use_remove_padding: true
+    lora_rank: 0
+    lora_alpha: 16
+    target_modules: all-linear
+    use_liger: false
+    use_fused_kernels: false
+    trust_remote_code: false
+  actor:
+    policy_loss: gspo_turn
+    enable_entropy_balanced_clipping: false
+    enable_entropy_balanced_advantage: false
+    strategy: fsdp
+    ppo_mini_batch_size: 8
+    ppo_micro_batch_size: null
+    ppo_micro_batch_size_per_gpu: null
+    use_dynamic_bsz: true
+    ppo_max_token_len_per_gpu: 16384
+    grad_clip: 1.0
+    clip_ratio: 0.2
+    clip_ratio_low: 0.003
+    clip_ratio_high: 0.004
+    clip_ratio_c: 3.0
+    loss_agg_mode: token-mean
+    entropy_coeff: 0
+    use_kl_loss: true
+    use_torch_compile: true
+    kl_loss_coef: 0.0
+    kl_loss_type: low_var_kl
+    ppo_epochs: 1
+    shuffle: false
+    ulysses_sequence_parallel_size: 1
+    checkpoint:
+      contents:
+      - model
+      - optimizer
+      - extra
+      save_contents:
+      - hf_model
+      - extra
+    optim:
+      lr: 1.0e-06
+      lr_warmup_steps: -1
+      lr_warmup_steps_ratio: 0.0
+      min_lr_ratio: 0.0
+      num_cycles: 0.5
+      warmup_style: constant
+      total_training_steps: -1
+      weight_decay: 0.01
+    fsdp_config:
+      wrap_policy:
+        min_num_params: 0
+      param_offload: false
+      optimizer_offload: false
+      offload_policy: false
+      reshard_after_forward: true
+      fsdp_size: -1
+  ref:
+    strategy: fsdp
+    fsdp_config:
+      param_offload: true
+      reshard_after_forward: true
+      wrap_policy:
+        min_num_params: 0
+    use_torch_compile: ${actor_rollout_ref.actor.use_torch_compile}
+    log_prob_micro_batch_size: null
+    log_prob_micro_batch_size_per_gpu: null
+    log_prob_use_dynamic_bsz: ${actor_rollout_ref.actor.use_dynamic_bsz}
+    log_prob_max_token_len_per_gpu: 32768
+    ulysses_sequence_parallel_size: ${actor_rollout_ref.actor.ulysses_sequence_parallel_size}
+  rollout:
+    name: vllm
+    mode: sync_with_tool_tree
+    chat_scheduler: null
+    temperature: 1.0
+    top_k: -1
+    top_p: 1
+    use_fire_sampling: false
+    prompt_length: ${data.max_prompt_length}
+    response_length: ${data.max_response_length}
+    initial_rollouts: 10
+    expansion_mode: entropy
+    expansion_iterations: 2
+    beam_size: 6
+    samples_per_tree: 22
+    data: ${data}
+    use_kl_in_reward: ${algorithm.use_kl_in_reward}
+    kl_penalty: ${algorithm.kl_penalty}
+    adv_estimator: ${algorithm.adv_estimator}
+    gamma: ${algorithm.gamma}
+    lam: ${algorithm.lam}
+    norm_adv_by_std_in_grpo: ${algorithm.norm_adv_by_std_in_grpo}
+    kl_ctrl: ${algorithm.kl_ctrl}
+    reward_model: ${reward_model}
+    custom_reward_function: ${custom_reward_function}
+    leaf_value_norm: true
+    node_value_mode: child_softmax
+    node_adv_mode: node_value
+    entropy_mixing_method: no_entropy
+    dtype: bfloat16
+    gpu_memory_utilization: 0.6
+    ignore_eos: false
+    enforce_eager: true
+    free_cache_engine: true
+    load_format: dummy_dtensor
+    layered_summon: false
+    tensor_model_parallel_size: 1
+    max_num_batched_tokens: 25000
+    max_model_len: null
+    max_num_seqs: 1024
+    log_prob_micro_batch_size: null
+    log_prob_micro_batch_size_per_gpu: null
+    log_prob_use_dynamic_bsz: ${actor_rollout_ref.actor.use_dynamic_bsz}
+    log_prob_max_token_len_per_gpu: 32768
+    disable_log_stats: true
+    enable_chunked_prefill: true
+    do_sample: true
+    'n': 22
+    branch_probability: 0.5
+    entropy_weight: 0.2
+    engine_kwargs:
+      vllm:
+        swap_space: null
+      sglang:
+        attention_backend: null
+    val_kwargs:
+      top_k: -1
+      top_p: 1.0
+      temperature: 0
+      'n': 1
+      do_sample: false
+    multi_turn:
+      enable: true
+      max_turns: null
+      tool_config_path: /scratch/user/sushil22_tamu.edu/projects/ATPO/ATPO/verl_atpo/examples/sglang_multiturn/config/tool_config/search_tool_config.yaml
+      format: chatml
+    enable_dynamic_rollouts: false
+    tools:
+      call_limit: 6
+      max_workers: 32
+      timeout: 120
+      retry_count: 4
+      verbose_logging: true
+      fail_on_error: true
+      tool_instances:
+        python:
+          class_path: verl.workers.rollout.tools.python_tool.PythonTool
+          params:
+            conda_path: <your_conda_path>
+            conda_env: verl
+        search:
+          class_path: verl.workers.rollout.tools.search_tool.BingSearchTool
+          params:
+            api_key: unused_local_rag_server
+            zone: serp_api_test
+            max_results: 10
+            result_length: 1000
+            location: cn
+            cache_file: /scratch/user/sushil22_tamu.edu/projects/ATPO/search_cache/search_cache.json
+            async_cache_write: true
+critic:
+  rollout_n: ${actor_rollout_ref.rollout.n}
+  strategy: fsdp
+  optim:
+    lr: 1.0e-05
+    lr_warmup_steps_ratio: 0.0
+    min_lr_ratio: null
+    warmup_style: constant
+    total_training_steps: -1
+    weight_decay: 0.01
+  model:
+    path: ~/models/deepseek-llm-7b-chat
+    use_shm: false
+    tokenizer_path: ${actor_rollout_ref.model.path}
+    override_config: {}
+    external_lib: ${actor_rollout_ref.model.external_lib}
+    enable_gradient_checkpointing: true
+    enable_activation_offload: false
+    use_remove_padding: false
+    trust_remote_code: ${actor_rollout_ref.model.trust_remote_code}
+    fsdp_config:
+      param_offload: false
+      optimizer_offload: false
+      offload_policy: false
+      reshard_after_forward: true
+      wrap_policy:
+        min_num_params: 0
+      fsdp_size: -1
+    lora_rank: 0
+    lora_alpha: 16
+    target_modules: all-linear
+  ppo_mini_batch_size: ${actor_rollout_ref.actor.ppo_mini_batch_size}
+  ppo_micro_batch_size: null
+  ppo_micro_batch_size_per_gpu: null
+  forward_micro_batch_size: ${critic.ppo_micro_batch_size}
+  forward_micro_batch_size_per_gpu: ${critic.ppo_micro_batch_size_per_gpu}
+  use_dynamic_bsz: ${actor_rollout_ref.actor.use_dynamic_bsz}
+  ppo_max_token_len_per_gpu: 32768
+  forward_max_token_len_per_gpu: ${critic.ppo_max_token_len_per_gpu}
+  ulysses_sequence_parallel_size: 1
+  ppo_epochs: ${actor_rollout_ref.actor.ppo_epochs}
+  shuffle: ${actor_rollout_ref.actor.shuffle}
+  grad_clip: 1.0
+  cliprange_value: 0.5
+  loss_agg_mode: ${actor_rollout_ref.actor.loss_agg_mode}
+  checkpoint:
+    contents:
+    - model
+    - optimizer
+    - extra
+reward_model:
+  enable: false
+  strategy: fsdp
+  model:
+    input_tokenizer: ${actor_rollout_ref.model.path}
+    path: ~/models/FsfairX-LLaMA3-RM-v0.1
+    use_shm: false
+    external_lib: ${actor_rollout_ref.model.external_lib}
+    use_remove_padding: false
+    use_fused_kernels: ${actor_rollout_ref.model.use_fused_kernels}
+    trust_remote_code: false
+    fsdp_config:
+      wrap_policy:
+        min_num_params: 0
+      param_offload: false
+      reshard_after_forward: true
+      fsdp_size: -1
+  micro_batch_size: null
+  micro_batch_size_per_gpu: null
+  max_length: null
+  ulysses_sequence_parallel_size: 1
+  use_dynamic_bsz: ${critic.use_dynamic_bsz}
+  forward_max_token_len_per_gpu: ${critic.forward_max_token_len_per_gpu}
+  reward_manager: naive
+  launch_reward_fn_async: false
+  sandbox_fusion:
+    url: null
+    max_concurrent: 64
+custom_reward_function:
+  path: /scratch/user/sushil22_tamu.edu/projects/ATPO/ATPO/verl_atpo/verl/utils/reward_score/deep_research_em.py
+  name: compute_score
+trainer:
+  balance_batch: true
+  total_epochs: 1
+  total_training_steps: null
+  project_name: Neurips-final-runs
+  experiment_name: atpo_hotpotqa_qwen8b
+  logger:
+  - console
+  - wandb
+  log_val_generations: 0
+  rollout_data_dir: /scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen8b/20260506_124901//rollout
+  validation_data_dir: /scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen8b/20260506_124901//validation
+  nnodes: 1
+  n_gpus_per_node: 8
+  save_freq: 50
+  resume_mode: resume_path
+  resume_from_path: /scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo-qwen3-8b-hotpotqa-step200/global_step_200
+  val_before_train: true
+  test_freq: 10
+  critic_warmup: 0
+  default_hdfs_dir: null
+  del_local_ckpt_after_load: false
+  default_local_dir: /scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen8b/20260506_124901/
+  max_actor_ckpt_to_keep: 2
+  max_critic_ckpt_to_keep: null
+  ray_wait_register_center_timeout: 300
+  device: cuda
+ray_init:
+  num_cpus: null

atpo_hotpotqa_qwen8b/20260506_124901/outputs/.hydra/hydra.yaml

@@ -0,0 +1,226 @@
+hydra:
+  run:
+    dir: /scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen8b/20260506_124901//outputs
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][HYDRA] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    loggers:
+      logging_example:
+        level: DEBUG
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: RUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.run.dir=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen8b/20260506_124901//outputs
+    - hydra.mode=RUN
+    task:
+    - algorithm.adv_estimator=grpo
+    - algorithm.kl_ctrl.kl_coef=0.0
+    - data.train_files=/scratch/user/sushil22_tamu.edu/projects/ATPO/rl_datasets/hotpotqa/train.parquet
+    - data.val_files=["/scratch/user/sushil22_tamu.edu/projects/ATPO/rl_datasets/hotpotqa/test_512.parquet"]
+    - data.prompt_key=prompt
+    - data.train_batch_size=64
+    - data.max_prompt_length=2000
+    - data.max_response_length=6192
+    - actor_rollout_ref.model.path=/scratch/user/sushil22_tamu.edu/projects/ATPO/models/Qwen3-8B
+    - actor_rollout_ref.model.enable_gradient_checkpointing=True
+    - actor_rollout_ref.model.use_remove_padding=True
+    - actor_rollout_ref.actor.policy_loss=gspo_turn
+    - actor_rollout_ref.actor.clip_ratio_low=3e-3
+    - actor_rollout_ref.actor.clip_ratio_high=4e-3
+    - actor_rollout_ref.actor.enable_entropy_balanced_clipping=False
+    - actor_rollout_ref.actor.enable_entropy_balanced_advantage=False
+    - actor_rollout_ref.actor.optim.lr=1e-6
+    - actor_rollout_ref.actor.ppo_mini_batch_size=8
+    - actor_rollout_ref.actor.use_dynamic_bsz=True
+    - actor_rollout_ref.actor.ppo_max_token_len_per_gpu=16384
+    - actor_rollout_ref.actor.use_kl_loss=True
+    - actor_rollout_ref.actor.kl_loss_coef=0.0
+    - actor_rollout_ref.actor.kl_loss_type=low_var_kl
+    - actor_rollout_ref.actor.fsdp_config.param_offload=False
+    - actor_rollout_ref.actor.fsdp_config.optimizer_offload=False
+    - actor_rollout_ref.rollout.enable_dynamic_rollouts=False
+    - actor_rollout_ref.rollout.log_prob_max_token_len_per_gpu=32768
+    - actor_rollout_ref.rollout.tensor_model_parallel_size=1
+    - actor_rollout_ref.rollout.name=vllm
+    - actor_rollout_ref.rollout.mode=sync_with_tool_tree
+    - actor_rollout_ref.rollout.gpu_memory_utilization=0.6
+    - actor_rollout_ref.rollout.n=22
+    - actor_rollout_ref.rollout.initial_rollouts=10
+    - actor_rollout_ref.rollout.expansion_mode=entropy
+    - actor_rollout_ref.rollout.expansion_iterations=2
+    - actor_rollout_ref.rollout.beam_size=6
+    - actor_rollout_ref.rollout.samples_per_tree=22
+    - actor_rollout_ref.rollout.branch_probability=0.5
+    - actor_rollout_ref.rollout.entropy_weight=0.2
+    - actor_rollout_ref.rollout.leaf_value_norm=True
+    - actor_rollout_ref.rollout.node_value_mode=child_softmax
+    - actor_rollout_ref.rollout.node_adv_mode=node_value
+    - actor_rollout_ref.rollout.tools.call_limit=6
+    - ++actor_rollout_ref.rollout.tools.tool_instances.search.params.cache_file=/scratch/user/sushil22_tamu.edu/projects/ATPO/search_cache/search_cache.json
+    - ++actor_rollout_ref.rollout.tools.tool_instances.search.params.api_key=unused_local_rag_server
+    - actor_rollout_ref.rollout.multi_turn.enable=True
+    - actor_rollout_ref.rollout.multi_turn.tool_config_path=/scratch/user/sushil22_tamu.edu/projects/ATPO/ATPO/verl_atpo/examples/sglang_multiturn/config/tool_config/search_tool_config.yaml
+    - actor_rollout_ref.ref.log_prob_max_token_len_per_gpu=32768
+    - actor_rollout_ref.ref.fsdp_config.param_offload=True
+    - reward_model.reward_manager=naive
+    - custom_reward_function.path=/scratch/user/sushil22_tamu.edu/projects/ATPO/ATPO/verl_atpo/verl/utils/reward_score/deep_research_em.py
+    - custom_reward_function.name=compute_score
+    - trainer.critic_warmup=0
+    - trainer.logger=[console, wandb]
+    - trainer.project_name=Neurips-final-runs
+    - trainer.experiment_name=atpo_hotpotqa_qwen8b
+    - trainer.n_gpus_per_node=8
+    - trainer.nnodes=1
+    - trainer.save_freq=50
+    - trainer.test_freq=10
+    - trainer.max_actor_ckpt_to_keep=2
+    - +actor_rollout_ref.actor.checkpoint.save_contents=[hf_model,extra]
+    - trainer.total_epochs=1
+    - trainer.default_local_dir=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen8b/20260506_124901/
+    - trainer.val_before_train=True
+    - trainer.rollout_data_dir=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen8b/20260506_124901//rollout
+    - trainer.validation_data_dir=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen8b/20260506_124901//validation
+    - ray_init.num_cpus=null
+    - trainer.resume_mode=resume_path
+    - trainer.resume_from_path=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo-qwen3-8b-hotpotqa-step200/global_step_200
+  job:
+    name: main_ppo
+    chdir: null
+    override_dirname: ++actor_rollout_ref.rollout.tools.tool_instances.search.params.api_key=unused_local_rag_server,++actor_rollout_ref.rollout.tools.tool_instances.search.params.cache_file=/scratch/user/sushil22_tamu.edu/projects/ATPO/search_cache/search_cache.json,+actor_rollout_ref.actor.checkpoint.save_contents=[hf_model,extra],actor_rollout_ref.actor.clip_ratio_high=4e-3,actor_rollout_ref.actor.clip_ratio_low=3e-3,actor_rollout_ref.actor.enable_entropy_balanced_advantage=False,actor_rollout_ref.actor.enable_entropy_balanced_clipping=False,actor_rollout_ref.actor.fsdp_config.optimizer_offload=False,actor_rollout_ref.actor.fsdp_config.param_offload=False,actor_rollout_ref.actor.kl_loss_coef=0.0,actor_rollout_ref.actor.kl_loss_type=low_var_kl,actor_rollout_ref.actor.optim.lr=1e-6,actor_rollout_ref.actor.policy_loss=gspo_turn,actor_rollout_ref.actor.ppo_max_token_len_per_gpu=16384,actor_rollout_ref.actor.ppo_mini_batch_size=8,actor_rollout_ref.actor.use_dynamic_bsz=True,actor_rollout_ref.actor.use_kl_loss=True,actor_rollout_ref.model.enable_gradient_checkpointing=True,actor_rollout_ref.model.path=/scratch/user/sushil22_tamu.edu/projects/ATPO/models/Qwen3-8B,actor_rollout_ref.model.use_remove_padding=True,actor_rollout_ref.ref.fsdp_config.param_offload=True,actor_rollout_ref.ref.log_prob_max_token_len_per_gpu=32768,actor_rollout_ref.rollout.beam_size=6,actor_rollout_ref.rollout.branch_probability=0.5,actor_rollout_ref.rollout.enable_dynamic_rollouts=False,actor_rollout_ref.rollout.entropy_weight=0.2,actor_rollout_ref.rollout.expansion_iterations=2,actor_rollout_ref.rollout.expansion_mode=entropy,actor_rollout_ref.rollout.gpu_memory_utilization=0.6,actor_rollout_ref.rollout.initial_rollouts=10,actor_rollout_ref.rollout.leaf_value_norm=True,actor_rollout_ref.rollout.log_prob_max_token_len_per_gpu=32768,actor_rollout_ref.rollout.mode=sync_with_tool_tree,actor_rollout_ref.rollout.multi_turn.enable=True,actor_rollout_ref.rollout.multi_turn.tool_config_path=/scratch/user/sushil22_tamu.edu/projects/ATPO/ATPO/verl_atpo/examples/sglang_multiturn/config/tool_config/search_tool_config.yaml,actor_rollout_ref.rollout.n=22,actor_rollout_ref.rollout.name=vllm,actor_rollout_ref.rollout.node_adv_mode=node_value,actor_rollout_ref.rollout.node_value_mode=child_softmax,actor_rollout_ref.rollout.samples_per_tree=22,actor_rollout_ref.rollout.tensor_model_parallel_size=1,actor_rollout_ref.rollout.tools.call_limit=6,algorithm.adv_estimator=grpo,algorithm.kl_ctrl.kl_coef=0.0,custom_reward_function.name=compute_score,custom_reward_function.path=/scratch/user/sushil22_tamu.edu/projects/ATPO/ATPO/verl_atpo/verl/utils/reward_score/deep_research_em.py,data.max_prompt_length=2000,data.max_response_length=6192,data.prompt_key=prompt,data.train_batch_size=64,data.train_files=/scratch/user/sushil22_tamu.edu/projects/ATPO/rl_datasets/hotpotqa/train.parquet,data.val_files=["/scratch/user/sushil22_tamu.edu/projects/ATPO/rl_datasets/hotpotqa/test_512.parquet"],ray_init.num_cpus=null,reward_model.reward_manager=naive,trainer.critic_warmup=0,trainer.default_local_dir=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen8b/20260506_124901/,trainer.experiment_name=atpo_hotpotqa_qwen8b,trainer.logger=[console,
+      wandb],trainer.max_actor_ckpt_to_keep=2,trainer.n_gpus_per_node=8,trainer.nnodes=1,trainer.project_name=Neurips-final-runs,trainer.resume_from_path=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo-qwen3-8b-hotpotqa-step200/global_step_200,trainer.resume_mode=resume_path,trainer.rollout_data_dir=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen8b/20260506_124901//rollout,trainer.save_freq=50,trainer.test_freq=10,trainer.total_epochs=1,trainer.val_before_train=True,trainer.validation_data_dir=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen8b/20260506_124901//validation
+    id: ???
+    num: ???
+    config_name: ppo_trainer_dr.yaml
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /scratch/user/sushil22_tamu.edu/projects/ATPO
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /scratch/user/sushil22_tamu.edu/projects/ATPO/ATPO/scripts/config
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen8b/20260506_124901/outputs
+    choices:
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: default
+      hydra/hydra_logging: default
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

atpo_hotpotqa_qwen8b/20260506_124901/outputs/.hydra/overrides.yaml

@@ -0,0 +1,70 @@
+- algorithm.adv_estimator=grpo
+- algorithm.kl_ctrl.kl_coef=0.0
+- data.train_files=/scratch/user/sushil22_tamu.edu/projects/ATPO/rl_datasets/hotpotqa/train.parquet
+- data.val_files=["/scratch/user/sushil22_tamu.edu/projects/ATPO/rl_datasets/hotpotqa/test_512.parquet"]
+- data.prompt_key=prompt
+- data.train_batch_size=64
+- data.max_prompt_length=2000
+- data.max_response_length=6192
+- actor_rollout_ref.model.path=/scratch/user/sushil22_tamu.edu/projects/ATPO/models/Qwen3-8B
+- actor_rollout_ref.model.enable_gradient_checkpointing=True
+- actor_rollout_ref.model.use_remove_padding=True
+- actor_rollout_ref.actor.policy_loss=gspo_turn
+- actor_rollout_ref.actor.clip_ratio_low=3e-3
+- actor_rollout_ref.actor.clip_ratio_high=4e-3
+- actor_rollout_ref.actor.enable_entropy_balanced_clipping=False
+- actor_rollout_ref.actor.enable_entropy_balanced_advantage=False
+- actor_rollout_ref.actor.optim.lr=1e-6
+- actor_rollout_ref.actor.ppo_mini_batch_size=8
+- actor_rollout_ref.actor.use_dynamic_bsz=True
+- actor_rollout_ref.actor.ppo_max_token_len_per_gpu=16384
+- actor_rollout_ref.actor.use_kl_loss=True
+- actor_rollout_ref.actor.kl_loss_coef=0.0
+- actor_rollout_ref.actor.kl_loss_type=low_var_kl
+- actor_rollout_ref.actor.fsdp_config.param_offload=False
+- actor_rollout_ref.actor.fsdp_config.optimizer_offload=False
+- actor_rollout_ref.rollout.enable_dynamic_rollouts=False
+- actor_rollout_ref.rollout.log_prob_max_token_len_per_gpu=32768
+- actor_rollout_ref.rollout.tensor_model_parallel_size=1
+- actor_rollout_ref.rollout.name=vllm
+- actor_rollout_ref.rollout.mode=sync_with_tool_tree
+- actor_rollout_ref.rollout.gpu_memory_utilization=0.6
+- actor_rollout_ref.rollout.n=22
+- actor_rollout_ref.rollout.initial_rollouts=10
+- actor_rollout_ref.rollout.expansion_mode=entropy
+- actor_rollout_ref.rollout.expansion_iterations=2
+- actor_rollout_ref.rollout.beam_size=6
+- actor_rollout_ref.rollout.samples_per_tree=22
+- actor_rollout_ref.rollout.branch_probability=0.5
+- actor_rollout_ref.rollout.entropy_weight=0.2
+- actor_rollout_ref.rollout.leaf_value_norm=True
+- actor_rollout_ref.rollout.node_value_mode=child_softmax
+- actor_rollout_ref.rollout.node_adv_mode=node_value
+- actor_rollout_ref.rollout.tools.call_limit=6
+- ++actor_rollout_ref.rollout.tools.tool_instances.search.params.cache_file=/scratch/user/sushil22_tamu.edu/projects/ATPO/search_cache/search_cache.json
+- ++actor_rollout_ref.rollout.tools.tool_instances.search.params.api_key=unused_local_rag_server
+- actor_rollout_ref.rollout.multi_turn.enable=True
+- actor_rollout_ref.rollout.multi_turn.tool_config_path=/scratch/user/sushil22_tamu.edu/projects/ATPO/ATPO/verl_atpo/examples/sglang_multiturn/config/tool_config/search_tool_config.yaml
+- actor_rollout_ref.ref.log_prob_max_token_len_per_gpu=32768
+- actor_rollout_ref.ref.fsdp_config.param_offload=True
+- reward_model.reward_manager=naive
+- custom_reward_function.path=/scratch/user/sushil22_tamu.edu/projects/ATPO/ATPO/verl_atpo/verl/utils/reward_score/deep_research_em.py
+- custom_reward_function.name=compute_score
+- trainer.critic_warmup=0
+- trainer.logger=[console, wandb]
+- trainer.project_name=Neurips-final-runs
+- trainer.experiment_name=atpo_hotpotqa_qwen8b
+- trainer.n_gpus_per_node=8
+- trainer.nnodes=1
+- trainer.save_freq=50
+- trainer.test_freq=10
+- trainer.max_actor_ckpt_to_keep=2
+- +actor_rollout_ref.actor.checkpoint.save_contents=[hf_model,extra]
+- trainer.total_epochs=1
+- trainer.default_local_dir=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen8b/20260506_124901/
+- trainer.val_before_train=True
+- trainer.rollout_data_dir=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen8b/20260506_124901//rollout
+- trainer.validation_data_dir=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo_hotpotqa_qwen8b/20260506_124901//validation
+- ray_init.num_cpus=null
+- trainer.resume_mode=resume_path
+- trainer.resume_from_path=/scratch/user/sushil22_tamu.edu/projects/ATPO/logs/atpo-qwen3-8b-hotpotqa-step200/global_step_200

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/global_step_200/actor/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/global_step_200/actor/config.json

@@ -0,0 +1,31 @@
+{
+  "_attn_implementation_autoset": true,
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 2560,
+  "initializer_range": 0.02,
+  "intermediate_size": 9728,
+  "max_position_embeddings": 40960,
+  "max_window_layers": 36,
+  "model_type": "qwen3",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 36,
+  "num_key_value_heads": 8,
+  "pad_token_id": 151643,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.51.3",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/global_step_200/actor/generation_config.json

@@ -0,0 +1,13 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "temperature": 0.6,
+  "top_k": 20,
+  "top_p": 0.95,
+  "transformers_version": "4.51.3"
+}

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/global_step_200/actor/model_world_size_8_rank_0.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c5ca219f75b95e92f00088168cc9a8b1ecff09d05f122ddca062e9bee78335f
+size 2205945402

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/global_step_200/actor/model_world_size_8_rank_2.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a2e37f19499645c737402a07765e97b5cac91a413d916bb659af377af7d4dc7
+size 2205945402

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/global_step_200/actor/model_world_size_8_rank_4.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42e0a4b239b283753df4f08d58ce1c7c662afabc2f335f63972133c2f2ec80b3
+size 2205945402

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/global_step_200/actor/model_world_size_8_rank_6.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:248c550fde02cec22aabedbf35dea04f401ee8b6c238d3b84273d55b7c7f5d6e
+size 2205945402

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/global_step_200/actor/optim_world_size_8_rank_3.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8cf1ae57e0d938f2a1fc6616eb9376f7c8db43f781ab88499ac538856e075db9
+size 4022501290

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/global_step_200/actor/optim_world_size_8_rank_7.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9652262a8e31a43f2b086d4304d48a90aadbc67dfbd182bd31b871aa3ca53e6
+size 4022501290

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/global_step_200/actor/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/global_step_200/actor/tokenizer_config.json

@@ -0,0 +1,240 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0].role == 'system' %}\n        {{- messages[0].content + '\\n\\n' }}\n    {%- endif %}\n    {{- \"# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0].role == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0].content + '<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}\n{%- for message in messages[::-1] %}\n    {%- set index = (messages|length - 1) - loop.index0 %}\n    {%- if ns.multi_step_tool and message.role == \"user\" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}\n        {%- set ns.multi_step_tool = false %}\n        {%- set ns.last_query_index = index %}\n    {%- endif %}\n{%- endfor %}\n{%- for message in messages %}\n    {%- if message.content is string %}\n        {%- set content = message.content %}\n    {%- else %}\n        {%- set content = '' %}\n    {%- endif %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) %}\n        {{- '<|im_start|>' + message.role + '\\n' + content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {%- set reasoning_content = '' %}\n        {%- if message.reasoning_content is string %}\n            {%- set reasoning_content = message.reasoning_content %}\n        {%- else %}\n            {%- if '</think>' in content %}\n                {%- set reasoning_content = content.split('</think>')[0].rstrip('\\n').split('<think>')[-1].lstrip('\\n') %}\n                {%- set content = content.split('</think>')[-1].lstrip('\\n') %}\n            {%- endif %}\n        {%- endif %}\n        {%- if loop.index0 > ns.last_query_index %}\n            {%- if loop.last or (not loop.last and reasoning_content) %}\n                {{- '<|im_start|>' + message.role + '\\n<think>\\n' + reasoning_content.strip('\\n') + '\\n</think>\\n\\n' + content.lstrip('\\n') }}\n            {%- else %}\n                {{- '<|im_start|>' + message.role + '\\n' + content }}\n            {%- endif %}\n        {%- else %}\n            {{- '<|im_start|>' + message.role + '\\n' + content }}\n        {%- endif %}\n        {%- if message.tool_calls %}\n            {%- for tool_call in message.tool_calls %}\n                {%- if (loop.first and content) or (not loop.first) %}\n                    {{- '\\n' }}\n                {%- endif %}\n                {%- if tool_call.function %}\n                    {%- set tool_call = tool_call.function %}\n                {%- endif %}\n                {{- '<tool_call>\\n{\"name\": \"' }}\n                {{- tool_call.name }}\n                {{- '\", \"arguments\": ' }}\n                {%- if tool_call.arguments is string %}\n                    {{- tool_call.arguments }}\n                {%- else %}\n                    {{- tool_call.arguments | tojson }}\n                {%- endif %}\n                {{- '}\\n</tool_call>' }}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if loop.first or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n    {%- if enable_thinking is defined and enable_thinking is false %}\n        {{- '<think>\\n\\n</think>\\n\\n' }}\n    {%- endif %}\n{%- endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/latest_checkpointed_iteration.txt

@@ -0,0 +1 @@
+200

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/1.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4505874d0421dddd4193dea8d38d4f0e511d577ab39dea751bbec47e17eb0b54
+size 17646341

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/10.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2e466b445b49996ea06998a90b1fd255a37cb9eda735d38acf829e8ec797e1c
+size 15670098

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/101.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0cc3d6ad3c589cabf1dbec218cb198a0acbd3e493d861cfbadcdd46229780d0
+size 13456182

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/118.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99083b5b83e00cc943dd5a7f0f03192be5e78864d97d0559c5ae2a514e58bfe3
+size 14548777

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

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/129.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a550c99130a9064c8a6cbabc943a96646b731ba5f81058aa1ac1f7013ad0251e
+size 14370467

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/13.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1ae25147f6ed86bf621fcdb7252274253f2632f0222feb72ad51a73379ad7607
+size 15077323

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/130.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd1f06ed8a8dec0e76ccd7c844cd52da123075ebfe8d74701f39dd74c3bc348d
+size 14168128

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/136.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce61183d07f48923a8bcb12de2e3a434b9129fe7d1845c5580746c8ef0881eb0
+size 14347063

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/143.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b27889e47f8d82bbd0706a4e86ecfd25134fee0f6b3b9461b7be85a42b594b0
+size 13529441

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/150.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f0292eb297b596f0250c2bd7c91ed1bacb732f3e481dc3f87a21777305ccb2e
+size 13971928

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/156.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79c049a6800e5807b01cda1d065331cc776cac6d2367ef531f8640d880235bc7
+size 14061460

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/157.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27d1fdd87d9e72045d5efb7f460cfbcd2c2880e0de5e758f179fa2136eaac769
+size 14182823

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/160.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b15e2cc4c5e5a84395833fdb5f76f63e3090465143f5bdadd44a16ee3af79a5f
+size 14120339

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/163.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f3790ab394e31b6e99d2ad1ddabe9e8d5acfdd8e5b662e2362a97868d45b7840
+size 13695941

cache_hotpotqa_qwen4b_ablation_noadvantage/20260504_231531/rollout/176.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b910d7159a01c6ab4b6f749412e4b20de38561d2ecd0f18eb1c3b48843157c9
+size 14044003