Add files using upload-large-folder tool

.gitattributes

@@ -650,3 +650,12 @@ wandb/run-20260622_183609-bhcblf16/run-bhcblf16.wandb filter=lfs diff=lfs merge=
 wandb/run-20260622_183641-19mq6q1z/run-19mq6q1z.wandb filter=lfs diff=lfs merge=lfs -text
 wandb/run-20251217_181044-gmt9ya4m/run-gmt9ya4m.wandb filter=lfs diff=lfs merge=lfs -text
 wandb/run-20251218_002911-rvpar64n/run-rvpar64n.wandb filter=lfs diff=lfs merge=lfs -text
+wandb/run-20251218_182448-on6j4c96/run-on6j4c96.wandb filter=lfs diff=lfs merge=lfs -text
+wandb/run-20251218_182745-ahjkdup0/run-ahjkdup0.wandb filter=lfs diff=lfs merge=lfs -text
+wandb/run-20251218_183419-wvypkl8w/run-wvypkl8w.wandb filter=lfs diff=lfs merge=lfs -text
+wandb/run-20251218_184253-g8kppnjh/run-g8kppnjh.wandb filter=lfs diff=lfs merge=lfs -text
+wandb/run-20251218_185248-ts36d12x/run-ts36d12x.wandb filter=lfs diff=lfs merge=lfs -text
+wandb/run-20251218_190243-6eiq1qnz/run-6eiq1qnz.wandb filter=lfs diff=lfs merge=lfs -text
+wandb/run-20251218_201054-j0xb9suw/run-j0xb9suw.wandb filter=lfs diff=lfs merge=lfs -text
+wandb/run-20251218_210047-o51cj6g5/run-o51cj6g5.wandb filter=lfs diff=lfs merge=lfs -text
+wandb/run-20251218_211229-yi7iasga/run-yi7iasga.wandb filter=lfs diff=lfs merge=lfs -text

wandb/run-20251218_182331-3lp68bhy/files/code/cleanrl/cleanrl/ppo_rubikscube.py

@@ -0,0 +1,517 @@
+# PPO with small MLP for RAGEN Rubik's Cube 2x2 using the existing env (no env edits)
+import os
+import random
+import time
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Tuple, Dict, Any, List
+import json
+import re
+
+import gymnasium as gym
+import numpy as np
+import torch
+import torch.nn as nn
+import torch.optim as optim
+import tyro
+from torch.distributions.categorical import Categorical
+
+import sys
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), '../../'))
+
+from ragen.env.rubikscube.env import RubiksCube2x2Env
+from ragen.env.rubikscube.config import RubiksCube2x2Config
+
+
+class RubiksCubeWrapper(gym.Env):
+    """
+    Adapter to use ragen RubiksCube2x2Env with Gymnasium vector API.
+    - Converts text observation to one-hot vector of 24 stickers x 6 colors.
+    - Maps agent actions [0..11] to env actions [1..12].
+    - Exposes proper observation_space and action_space.
+    """
+    metadata = {"render_modes": ["rgb_array", "human", "ansi"]}
+
+    def __init__(self, env: RubiksCube2x2Env):
+        super().__init__()
+        self._env = env
+        # 24 stickers, 6 colors -> one-hot size 144
+        self._colors = ['W', 'O', 'G', 'R', 'B', 'Y']
+        self._color_to_idx = {c: i for i, c in enumerate(self._colors)}
+        self.observation_space = gym.spaces.Box(low=0.0, high=1.0, shape=(24 * len(self._colors),), dtype=np.float32)
+        self.action_space = gym.spaces.Discrete(12)
+        # precompile regex to extract 4 letters within [X, X]\n [X, X]
+        # Lines look like: "Up (U):    [W, W]\n       [W, W]"
+        self._face_pat = re.compile(r"\[(?:\s*([A-Z])\s*,\s*([A-Z])\s*\])\n\s*\[(?:\s*([A-Z])\s*,\s*([A-Z])\s*\])")
+
+    def _encode_obs(self, text_obs: str) -> np.ndarray:
+        # Extract the six face blocks in the order U, L, F, R, B, D as rendered
+        faces_order = ["Up (U):", "Left (L):", "Front (F):", "Right (R):", "Back (B):", "Down (D):"]
+        onehots: List[int] = []
+        # Build a map from header to its block text
+        lines = text_obs.splitlines()
+        # Collect blocks starting after header line and including the next line for second row
+        i = 0
+        blocks: List[str] = []
+        while i < len(lines):
+            line = lines[i]
+            for header in faces_order:
+                if line.startswith(header):
+                    # Join current line's bracketed pair and the next line (which contains the second pair)
+                    # Remove the "Header:" prefix to keep only the bracket portions
+                    content = line[len(header):].strip()
+                    next_line = lines[i + 1] if i + 1 < len(lines) else ""
+                    block = f"{content}\n{next_line}"
+                    blocks.append(block)
+                    break
+            i += 1
+        # Fallback: if regex fails, return zeros
+        if len(blocks) != 6:
+            return np.zeros(24 * len(self._colors), dtype=np.float32)
+        stickers: List[int] = []
+        for blk in blocks:
+            m = self._face_pat.search(blk)
+            if not m:
+                return np.zeros(24 * len(self._colors), dtype=np.float32)
+            # order: 0,1,2,3 per render() doc
+            c0, c1, c2, c3 = m.group(1), m.group(2), m.group(3), m.group(4)
+            stickers.extend([c0, c1, c2, c3])
+        # stickers now length 24; convert to one-hot
+        grid = np.zeros((24, len(self._colors)), dtype=np.float32)
+        for idx, ch in enumerate(stickers):
+            cidx = self._color_to_idx.get(ch, None)
+            if cidx is not None:
+                grid[idx, cidx] = 1.0
+        return grid.reshape(-1)
+
+    def reset(self, *, seed: int | None = None, options: Dict[str, Any] | None = None):
+        text_obs = self._env.reset(seed=seed)
+        obs = self._encode_obs(text_obs)
+        return obs, {}
+
+    def step(self, action: int):
+        mapped = int(action) + 1  # 0..11 -> 1..12
+        text_obs, reward, done, info = self._env.step(mapped)
+        obs = self._encode_obs(text_obs)
+        terminated = bool(done)
+        truncated = False
+        return obs, float(reward), terminated, truncated, info or {}
+
+    def render(self):
+        return self._env.render()
+
+    def close(self):
+        self._env.close()
+
+
+@dataclass
+class Args:
+    exp_name: str = os.path.basename(__file__)[: -len(".py")]
+    seed: int = 1
+    torch_deterministic: bool = True
+    cuda: bool = True
+    track: bool = True
+    wandb_project_name: str = "cleanRL"
+    wandb_entity: str | None = None
+    capture_video: bool = False
+
+    # Algorithm
+    env_id: str = "RubiksCube2x2"
+    total_timesteps: int = 1000_000
+    learning_rate: float = 2.5e-4
+    num_envs: int = 8
+    num_steps: int = 128
+    anneal_lr: bool = True
+    gamma: float = 0.99
+    gae_lambda: float = 0.95
+    num_minibatches: int = 4
+    update_epochs: int = 4
+    norm_adv: bool = True
+    clip_coef: float = 0.2
+    clip_vloss: bool = True
+    ent_coef: float = 0.01
+    vf_coef: float = 0.5
+    max_grad_norm: float = 0.5
+    target_kl: float | None = None
+
+    # Rubik specific
+    scramble_depth: int = 1
+    max_steps_env: int = 20
+
+    # runtime filled
+    batch_size: int = 0
+    minibatch_size: int = 0
+    num_iterations: int = 0
+    eval_splits: int = 2
+    eval_episodes: int = 4000
+
+
+def make_env(idx, run_name, seed, scramble_depth, max_steps_env, capture_video=False):
+    def thunk():
+        config = RubiksCube2x2Config(scramble_depth=scramble_depth, max_steps=max_steps_env, render_mode='text')
+        env = RubiksCube2x2Env(config)
+        env = RubiksCubeWrapper(env)
+        env = gym.wrappers.TimeLimit(env, max_episode_steps=max_steps_env)
+        env = gym.wrappers.RecordEpisodeStatistics(env)
+        if capture_video and idx == 0:
+            env = gym.wrappers.RecordVideo(env, f"videos/{run_name}")
+        return env
+    return thunk
+
+
+def layer_init(layer, std=np.sqrt(2), bias_const=0.0):
+    torch.nn.init.orthogonal_(layer.weight, std)
+    torch.nn.init.constant_(layer.bias, bias_const)
+    return layer
+
+
+class Agent(nn.Module):
+    def __init__(self, envs):
+        super().__init__()
+        obs_shape = int(np.array(envs.single_observation_space.shape).prod())
+        hidden = 128
+        self.critic = nn.Sequential(
+            layer_init(nn.Linear(obs_shape, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, 1), std=1.0),
+        )
+        self.actor = nn.Sequential(
+            layer_init(nn.Linear(obs_shape, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, envs.single_action_space.n), std=0.01),
+        )
+
+    def get_value(self, x):
+        return self.critic(x)
+
+    def get_action_and_value(self, x, action=None):
+        logits = self.actor(x)
+        probs = Categorical(logits=logits)
+        if action is None:
+            action = probs.sample()
+        return action, probs.log_prob(action), probs.entropy(), self.critic(x)
+
+
+if __name__ == "__main__":
+    args = tyro.cli(Args)
+    args.batch_size = int(args.num_envs * args.num_steps)
+    args.minibatch_size = int(args.batch_size // args.num_minibatches)
+    args.num_iterations = args.total_timesteps // args.batch_size
+    run_name = f"{args.env_id}__{args.exp_name}__{args.seed}__{int(time.time())}"
+
+    if args.track:
+        import wandb
+        wandb.init(
+            project=args.wandb_project_name,
+            entity=args.wandb_entity,
+            config=vars(args),
+            name=run_name,
+            monitor_gym=True,
+            save_code=True,
+        )
+        try:
+            wandb.define_metric("global_step")
+            for prefix in ["train/*", "rollout/*", "eval/*", "losses/*", "charts/*", "perf/*"]:
+                wandb.define_metric(prefix, step_metric="global_step")
+        except Exception:
+            pass
+
+    # seeding
+    random.seed(args.seed)
+    np.random.seed(args.seed)
+    torch.manual_seed(args.seed)
+    torch.backends.cudnn.deterministic = args.torch_deterministic
+
+    device = torch.device("cuda" if torch.cuda.is_available() and args.cuda else "cpu")
+
+    # envs
+    envs = gym.vector.SyncVectorEnv([
+        make_env(i, run_name, args.seed, args.scramble_depth, args.max_steps_env, args.capture_video)
+        for i in range(args.num_envs)
+    ])
+    assert isinstance(envs.single_action_space, gym.spaces.Discrete)
+
+    agent = Agent(envs).to(device)
+    optimizer = optim.Adam(agent.parameters(), lr=args.learning_rate, eps=1e-5)
+
+    # storage
+    obs = torch.zeros((args.num_steps, args.num_envs) + envs.single_observation_space.shape).to(device)
+    actions = torch.zeros((args.num_steps, args.num_envs) + envs.single_action_space.shape).to(device)
+    logprobs = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    rewards = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    dones = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    values = torch.zeros((args.num_steps, args.num_envs)).to(device)
+
+    # start
+    global_step = 0
+    start_time = time.time()
+    next_obs, _ = envs.reset(seed=args.seed)
+    next_obs = torch.Tensor(next_obs).to(device)
+    next_done = torch.zeros(args.num_envs).to(device)
+
+    # eval helper similar to FrozenLake: collect greedy eval trajectories and write metrics.json
+    def collect_eval_trajectories(agent_model, make_env_fn, n_episodes, step_tag):
+        out_dir = Path(f"runs/{run_name}/trajectories/step_{step_tag}")
+        out_dir.mkdir(parents=True, exist_ok=True)
+        out_path = out_dir / "trajectories.jsonl"
+        env = make_env_fn()
+        collected = 0
+        summary_returns = []
+        summary_success = []
+        with out_path.open("w") as f:
+            while collected < n_episodes:
+                state, _ = env.reset(seed=args.seed + 100000 + collected)
+                traj_states = [state.tolist()]
+                traj_actions = []
+                traj_rewards = []
+                traj_dones = []
+                traj_success = []
+                done = False
+                step_count = 0
+                max_eval_steps = getattr(env, '_max_episode_steps', None) or int(args.max_steps_env)
+                while not done:
+                    with torch.no_grad():
+                        logits = agent_model.actor(torch.tensor(state, dtype=torch.float32, device=device).unsqueeze(0))
+                        action = int(torch.argmax(logits, dim=1).item())
+                    next_state, reward, terminated, truncated, info = env.step(action)
+                    traj_actions.append(int(action))
+                    traj_rewards.append(float(reward))
+                    step_count += 1
+                    d = bool(terminated) or bool(truncated) or (step_count >= max_eval_steps)
+                    traj_dones.append(d)
+                    traj_success.append(bool((info or {}).get('success', False)))
+                    state = next_state
+                    traj_states.append(state.tolist())
+                    done = d
+                ep_ret = float(sum(traj_rewards))
+                ep_succ = bool(any(traj_success))
+                record = {
+                    "states": traj_states,
+                    "actions": traj_actions,
+                    "rewards": traj_rewards,
+                    "dones": traj_dones,
+                    "success": traj_success,
+                    "episode_return": ep_ret,
+                    "episode_success": ep_succ,
+                }
+                f.write(json.dumps(record) + "\n")
+                collected += 1
+                summary_returns.append(ep_ret)
+                summary_success.append(1.0 if ep_succ else 0.0)
+        env.close()
+        try:
+            metrics = {
+                "global_step": int(step_tag),
+                "episodes": int(n_episodes),
+                "success_rate": float(np.mean(summary_success)) if len(summary_success) else 0.0,
+                "avg_return": float(np.mean(summary_returns)) if len(summary_returns) else 0.0,
+                "std_return": float(np.std(summary_returns)) if len(summary_returns) else 0.0,
+            }
+            with (out_dir / "metrics.json").open("w") as mf:
+                json.dump(metrics, mf)
+        except Exception as e:
+            print(f"Warning: failed to write eval metrics: {e}")
+
+    # training loop
+    eval_every_iters = max(1, args.num_iterations // args.eval_splits)
+    for iteration in range(1, args.num_iterations + 1):
+        # Anneal LR
+        if args.anneal_lr:
+            frac = 1.0 - (iteration - 1.0) / args.num_iterations
+            lrnow = frac * args.learning_rate
+            optimizer.param_groups[0]["lr"] = lrnow
+
+        # accumulate per-iteration episode successes
+        iter_successes = []
+        for step in range(0, args.num_steps):
+            global_step += args.num_envs
+            obs[step] = next_obs
+            dones[step] = next_done
+
+            with torch.no_grad():
+                action, logprob, _, value = agent.get_action_and_value(next_obs)
+                values[step] = value.flatten()
+            actions[step] = action
+            logprobs[step] = logprob
+
+            next_obs, reward, terminations, truncations, infos = envs.step(action.cpu().numpy())
+            next_done = np.logical_or(terminations, truncations)
+            rewards[step] = torch.tensor(reward).to(device).view(-1)
+            next_obs, next_done = torch.Tensor(next_obs).to(device), torch.Tensor(next_done).to(device)
+
+            # Episode stats logging similar to FrozenLake
+            try:
+                mask = None
+                if isinstance(infos, dict):
+                    if "_episode" in infos:
+                        mask = np.asarray(infos["_episode"]).astype(bool)
+                    elif "episode" in infos and isinstance(infos["episode"], dict) and "_l" in infos["episode"]:
+                        mask = np.asarray(infos["episode"]["_l"]).astype(bool)
+                if mask is not None and np.any(mask):
+                    r_arr = np.asarray(infos.get("episode", {}).get("r", np.zeros_like(mask, dtype=float)))
+                    l_arr = np.asarray(infos.get("episode", {}).get("l", np.zeros_like(mask, dtype=int)))
+                    # prefer success from info; fallback to ep return > 0
+                    if "success" in infos:
+                        succ_arr = np.asarray(infos.get("success", np.zeros_like(mask, dtype=bool))).astype(float)
+                    else:
+                        try:
+                            succ_arr = (np.asarray(r_arr) > 0).astype(float)
+                        except Exception:
+                            succ_arr = np.zeros_like(mask, dtype=float)
+                    # collect iteration successes for training success rate
+                    try:
+                        for s in np.asarray(succ_arr)[mask]:
+                            iter_successes.append(float(s))
+                    except Exception:
+                        pass
+                    if args.track:
+                        try:
+                            import wandb
+                            log_dict = {
+                                "global_step": int(global_step),
+                                "rollout/ep_rew_mean": float(np.mean(r_arr[mask])) if np.any(mask) else None,
+                                "rollout/ep_len_mean": float(np.mean(l_arr[mask])) if np.any(mask) else None,
+                                "rollout/success_rate": float(np.mean(succ_arr[mask])) if np.any(mask) else None,
+                            }
+                            wandb.log(log_dict, step=global_step)
+                        except Exception:
+                            pass
+            except Exception:
+                pass
+
+        # GAE
+        with torch.no_grad():
+            next_value = agent.get_value(next_obs).reshape(1, -1)
+            advantages = torch.zeros_like(rewards).to(device)
+            lastgaelam = 0
+            for t in reversed(range(args.num_steps)):
+                if t == args.num_steps - 1:
+                    nextnonterminal = 1.0 - next_done
+                    nextvalues = next_value
+                else:
+                    nextnonterminal = 1.0 - dones[t + 1]
+                    nextvalues = values[t + 1]
+                delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
+                advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
+            returns = advantages + values
+
+        # flatten batch
+        b_obs = obs.reshape((-1,) + envs.single_observation_space.shape)
+        b_logprobs = logprobs.reshape(-1)
+        b_actions = actions.reshape((-1,) + envs.single_action_space.shape)
+        b_advantages = advantages.reshape(-1)
+        b_returns = returns.reshape(-1)
+        b_values = values.reshape(-1)
+
+        # update
+        b_inds = np.arange(args.batch_size)
+        clipfracs = []
+        for epoch in range(args.update_epochs):
+            np.random.shuffle(b_inds)
+            for start in range(0, args.batch_size, args.minibatch_size):
+                end = start + args.minibatch_size
+                mb_inds = b_inds[start:end]
+
+                _, newlogprob, entropy, newvalue = agent.get_action_and_value(b_obs[mb_inds], b_actions.long()[mb_inds])
+                logratio = newlogprob - b_logprobs[mb_inds]
+                ratio = logratio.exp()
+
+                with torch.no_grad():
+                    old_approx_kl = (-logratio).mean()
+                    approx_kl = ((ratio - 1) - logratio).mean()
+                    clipfracs += [((ratio - 1.0).abs() > args.clip_coef).float().mean().item()]
+
+                mb_advantages = b_advantages[mb_inds]
+                if args.norm_adv:
+                    mb_advantages = (mb_advantages - mb_advantages.mean()) / (mb_advantages.std() + 1e-8)
+
+                pg_loss1 = -mb_advantages * ratio
+                pg_loss2 = -mb_advantages * torch.clamp(ratio, 1 - args.clip_coef, 1 + args.clip_coef)
+                pg_loss = torch.max(pg_loss1, pg_loss2).mean()
+
+                newvalue = newvalue.view(-1)
+                if args.clip_vloss:
+                    v_loss_unclipped = (newvalue - b_returns[mb_inds]) ** 2
+                    v_clipped = b_values[mb_inds] + torch.clamp(
+                        newvalue - b_values[mb_inds], -args.clip_coef, args.clip_coef,
+                    )
+                    v_loss_clipped = (v_clipped - b_returns[mb_inds]) ** 2
+                    v_loss = 0.5 * torch.max(v_loss_unclipped, v_loss_clipped).mean()
+                else:
+                    v_loss = 0.5 * ((newvalue - b_returns[mb_inds]) ** 2).mean()
+
+                entropy_loss = entropy.mean()
+                loss = pg_loss - args.ent_coef * entropy_loss + v_loss * args.vf_coef
+
+                optimizer.zero_grad()
+                loss.backward()
+                nn.utils.clip_grad_norm_(agent.parameters(), args.max_grad_norm)
+                optimizer.step()
+
+            if args.target_kl is not None and approx_kl > args.target_kl:
+                break
+
+        # metrics
+        y_pred, y_true = b_values.cpu().numpy(), b_returns.cpu().numpy()
+        var_y = np.var(y_true)
+        explained_var = np.nan if var_y == 0 else 1 - np.var(y_true - y_pred) / var_y
+
+        sps = int(global_step / (time.time() - start_time))
+        train_success_rate = float(np.mean(iter_successes)) if len(iter_successes) else 0.0
+        print(f"Iter {iteration:4d}/{args.num_iterations} | SPS: {sps:5d} | R: {rewards.mean().item():6.3f}")
+        if args.track:
+            try:
+                import wandb
+                wandb.log({
+                    "global_step": int(global_step),
+                    "charts/progress": float(100.0 * iteration / max(1, args.num_iterations)),
+                    "train/value_loss": float(v_loss.item()),
+                    "train/policy_loss": float(pg_loss.item()),
+                    "losses/value_loss": float(v_loss.item()),
+                    "losses/policy_loss": float(pg_loss.item()),
+                    "train/entropy": float(entropy_loss.item()),
+                    "train/old_approx_kl": float(old_approx_kl.item()),
+                    "train/approx_kl": float(approx_kl.item()),
+                    "train/clipfrac": float(np.mean(clipfracs)) if len(clipfracs) else 0.0,
+                    "losses/explained_variance": float(explained_var),
+                    "charts/avg_reward": float(rewards.mean().item()),
+                    "charts/avg_value": float(values.mean().item()),
+                    "perf/SPS": int(sps),
+                    "charts/SPS": int(sps),
+                    "train/success_rate": train_success_rate,
+                    "charts/train_success_rate": train_success_rate,
+                    "train/learning_rate": float(optimizer.param_groups[0]["lr"]),
+                }, step=global_step)
+            except Exception:
+                pass
+
+        # periodic evaluation collection
+        if iteration % eval_every_iters == 0:
+            try:
+                eval_thunk = make_env(0, run_name, args.seed + 9999, args.scramble_depth, args.max_steps_env, False)
+                collect_eval_trajectories(agent, eval_thunk, n_episodes=args.eval_episodes, step_tag=global_step)
+                if args.track:
+                    try:
+                        import json as _json
+                        from pathlib import Path as _Path
+                        mpath = _Path(f"runs/{run_name}/trajectories/step_{global_step}/metrics.json")
+                        if mpath.exists():
+                            with mpath.open("r") as mf:
+                                metrics = _json.load(mf)
+                            wandb.log({
+                                "eval/success_rate": metrics.get("success_rate"),
+                                "eval/avg_return": metrics.get("avg_return"),
+                                "eval/std_return": metrics.get("std_return"),
+                                "eval/episodes": metrics.get("episodes"),
+                            }, step=global_step)
+                    except Exception:
+                        pass
+                print(f"Collected {args.eval_episodes} eval trajectories at global_step {global_step}")
+            except Exception as e:
+                print(f"Warning: eval trajectory collection failed at step {global_step}: {e}")
+
+    envs.close()

wandb/run-20251218_182331-3lp68bhy/files/requirements.txt

@@ -0,0 +1,315 @@
+colorama==0.4.6
+psutil==7.1.2
+ragen==0.1
+ragen==0.1
+setuptools==80.9.0
+wheel==0.45.1
+pip==25.3
+ragen==0.1
+pytz==2025.2
+nvidia-cusparselt-cu12==0.6.2
+mpmath==1.3.0
+typing_extensions==4.15.0
+sympy==1.13.1
+pylatexenc==2.10
+nvidia-cusolver-cu12==11.6.1.9
+numpy==1.26.4
+Markdown==3.10
+absl-py==2.3.1
+pyasn1_modules==0.4.2
+Werkzeug==3.1.3
+networkx==3.3
+MarkupSafe==2.1.5
+fsspec==2025.9.0
+filelock==3.19.1
+nvidia-cudnn-cu12==9.1.0.70
+Jinja2==3.1.6
+tensorboard==2.20.0
+triton==3.2.0
+pillow==11.3.0
+xxhash==3.6.0
+nvidia-nvtx-cu12==12.4.127
+nvidia-nvjitlink-cu12==12.4.127
+nvidia-nccl-cu12==2.21.5
+nvidia-curand-cu12==10.3.5.147
+wrapt==2.0.1
+nvidia-cufft-cu12==11.2.1.3
+pyjnius==1.7.0
+nvidia-cuda-runtime-cu12==12.4.127
+nvidia-cuda-nvrtc-cu12==12.4.127
+websockets==15.0.1
+nvidia-cuda-cupti-cu12==12.4.127
+zipp==3.23.0
+nvidia-cublas-cu12==12.4.5.8
+nvidia-cusparse-cu12==12.3.1.170
+flash_attn==2.8.3
+rsa==4.9.1
+einops==0.8.1
+py-cpuinfo==9.0.0
+pure_eval==0.2.3
+ptyprocess==0.7.0
+gym-notices==0.1.0
+flatbuffers==25.9.23
+fastrlock==0.8.3
+Farama-Notifications==0.0.4
+faiss-gpu==1.7.2
+antlr4-python3-runtime==4.9.3
+wcwidth==0.2.14
+wasabi==1.1.3
+uvloop==0.22.1
+urllib3==2.5.0
+tzdata==2025.2
+typing-inspection==0.4.2
+traitlets==5.14.3
+tqdm==4.67.1
+tomli==2.3.0
+threadpoolctl==3.6.0
+tenacity==9.1.2
+tabulate==0.9.0
+spacy-loggers==1.0.5
+spacy-legacy==3.0.12
+soupsieve==2.8
+sniffio==1.3.1
+smmap==5.0.2
+six==1.17.0
+shellingham==1.5.4
+sentencepiece==0.2.1
+scipy==1.15.3
+safetensors==0.6.2
+rpds-py==0.29.0
+rignore==0.7.6
+regex==2025.11.3
+RapidFuzz==3.14.3
+rank-bm25==0.2.2
+pyzmq==27.1.0
+PyYAML==6.0.3
+pytokens==0.3.0
+python-multipart==0.0.20
+python-json-logger==4.0.0
+python-dotenv==1.2.1
+PySocks==1.7.1
+pyparsing==3.2.5
+PyJWT==2.10.1
+Pygments==2.19.2
+pygame==2.6.1
+pydantic_core==2.41.5
+pycparser==2.23
+pycountry==24.6.1
+pybind11==3.0.1
+pyarrow==22.0.0
+psutil==7.1.3
+protobuf==4.25.8
+propcache==0.4.1
+prometheus_client==0.23.1
+platformdirs==4.5.0
+pexpect==4.9.0
+pathspec==0.12.1
+partial-json-parser==0.2.1.1.post6
+parso==0.8.5
+packaging==25.0
+orjson==3.11.4
+opentelemetry-semantic-conventions-ai==0.4.13
+opencv-python-headless==4.12.0.88
+ninja==1.13.0
+nest-asyncio==1.6.0
+mypy_extensions==1.1.0
+murmurhash==1.0.15
+multidict==6.7.0
+msgspec==0.19.0
+msgpack==1.1.2
+mdurl==0.1.2
+loguru==0.7.3
+llvmlite==0.44.0
+llguidance==0.7.30
+lark==1.2.2
+kiwisolver==1.4.9
+joblib==1.5.2
+jiter==0.12.0
+itsdangerous==2.2.0
+interegular==0.3.3
+ImageIO==2.37.2
+idna==3.11
+humanfriendly==10.0
+httpx-sse==0.4.3
+httptools==0.7.1
+html2text==2025.4.15
+hf-xet==1.2.0
+h11==0.16.0
+grpcio==1.76.0
+frozenlist==1.8.0
+fonttools==4.60.1
+executing==2.2.1
+exceptiongroup==1.3.0
+eval_type_backport==0.2.2
+docstring_parser==0.17.0
+dnspython==2.8.0
+distro==1.9.0
+diskcache==5.6.3
+dill==0.4.0
+decorator==5.2.1
+debugpy==1.8.17
+Cython==3.2.1
+cymem==2.0.13
+cycler==0.12.1
+cupy-cuda12x==13.6.0
+contourpy==1.3.2
+colorama==0.4.6
+codetiming==1.4.0
+cloudpickle==3.1.2
+cloudpathlib==0.23.0
+click==8.3.1
+charset-normalizer==3.4.4
+certifi==2025.11.12
+catalogue==2.0.10
+cachetools==6.2.2
+blis==1.3.2
+blinker==1.9.0
+blake3==1.0.8
+attrs==25.4.0
+async-timeout==5.0.1
+asttokens==3.0.1
+astor==0.8.1
+annotated-types==0.7.0
+annotated-doc==0.0.4
+airportsdata==20250909
+aiohappyeyeballs==2.6.1
+yarl==1.22.0
+uvicorn==0.38.0
+typer-slim==0.20.0
+thefuzz==0.22.1
+stack-data==0.6.3
+srsly==2.5.1
+smart_open==7.5.0
+sentry-sdk==2.44.0
+scikit-learn==1.7.2
+requests==2.32.5
+referencing==0.37.0
+python-dateutil==2.9.0.post0
+pydantic==2.12.4
+prompt_toolkit==3.0.52
+preshed==3.0.11
+opentelemetry-proto==1.26.0
+omegaconf==2.3.0
+numba==0.61.2
+nltk==3.9.2
+multiprocess==0.70.18
+matplotlib-inline==0.2.1
+markdown-it-py==4.0.0
+jedi==0.19.2
+importlib_metadata==8.0.0
+httpcore==1.0.9
+gymnasium==1.2.2
+gym==0.26.2
+googleapis-common-protos==1.72.0
+gitdb==4.0.12
+gguf==0.17.1
+Flask==3.1.2
+faiss-cpu==1.13.0
+email-validator==2.3.0
+depyf==0.18.0
+Deprecated==1.3.1
+coloredlogs==15.0.1
+cffi==2.0.0
+black==25.11.0
+beautifulsoup4==4.14.2
+anyio==4.11.0
+aiosignal==1.4.0
+watchfiles==1.1.1
+tiktoken==0.12.0
+starlette==0.49.3
+sse-starlette==3.0.3
+rich==14.2.0
+pydantic-settings==2.12.0
+pydantic-extra-types==2.10.6
+pandas==2.3.3
+opentelemetry-exporter-otlp-proto-common==1.26.0
+onnxruntime==1.23.2
+matplotlib==3.10.7
+lm-format-enforcer==0.10.12
+jsonschema-specifications==2025.9.1
+ipython==8.37.0
+hydra-core==1.3.2
+huggingface-hub==0.36.0
+httpx==0.28.1
+gym-sokoban==0.0.6
+GitPython==3.1.45
+cryptography==46.0.3
+confection==0.1.5
+cleantext==1.1.4
+aiohttp==3.13.2
+weasel==0.4.3
+wandb==0.23.0
+typer==0.19.2
+virtualenv==20.35.4
+pyasn1==0.6.1
+torch==2.6.0
+tokenizers==0.22.1
+thinc==8.3.9
+rich-toolkit==0.15.1
+prometheus-fastapi-instrumentator==7.1.0
+openai==2.8.0
+jsonschema==4.25.1
+gdown==5.2.0
+fastapi==0.121.2
+anthropic==0.73.0
+xformers==0.0.29.post2
+transformers==4.57.1
+proto-plus==1.26.1
+torchvision==0.21.0
+torchdata==0.11.0
+torchaudio==2.6.0
+together==1.5.30
+tensordict==0.8.3
+spacy==3.8.9
+ray==2.51.1
+outlines_core==0.1.26
+mistral_common==1.8.5
+mcp==1.21.1
+fastapi-cloud-cli==0.3.1
+fastapi-cli==0.0.16
+datasets==4.4.1
+accelerate==1.11.0
+xgrammar==0.1.18
+pyserini==1.2.0
+peft==0.18.0
+outlines==0.1.11
+opentelemetry-exporter-otlp-proto-http==1.26.0
+opentelemetry-exporter-otlp-proto-grpc==1.26.0
+kimina-client==0.2.1
+compressed-tensors==0.9.3
+opentelemetry-exporter-otlp==1.26.0
+vllm==0.8.5
+py-spy==0.4.1
+opencensus-context==0.1.3
+distlib==0.4.0
+colorful==0.5.8
+tensorboard-data-server==0.7.2
+google-auth==2.43.0
+google-api-core==2.28.1
+aiohttp-cors==0.8.1
+opentelemetry-exporter-prometheus==0.59b0
+opencensus==0.11.4
+verl==0.5.0.dev0
+opentelemetry-api==1.38.0
+opentelemetry-semantic-conventions==0.59b0
+opentelemetry-sdk==1.38.0
+typeguard==4.4.4
+shtab==1.8.0
+tyro==0.9.35
+verl==0.5.0.dev0
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2

wandb/run-20251218_182331-3lp68bhy/files/wandb-metadata.json

@@ -0,0 +1,92 @@
+{
+  "os": "Linux-5.14.0-284.25.1.el9_2.x86_64-x86_64-with-glibc2.35",
+  "python": "CPython 3.10.19",
+  "startedAt": "2025-12-18T10:23:31.950845Z",
+  "args": [
+    "--track"
+  ],
+  "program": "/mnt/general/wanghy/RAGEN/cleanrl/cleanrl/ppo_rubikscube.py",
+  "codePath": "cleanrl/cleanrl/ppo_rubikscube.py",
+  "codePathLocal": "cleanrl/cleanrl/ppo_rubikscube.py",
+  "git": {
+    "remote": "https://github.com/mll-lab-nu/RAGEN.git",
+    "commit": "8d73639b99b38265453f898b8d6af7d4af50d56e"
+  },
+  "email": "haoyu-wa22@mails.tsinghua.edu.cn",
+  "root": "/mnt/general/wanghy/RAGEN",
+  "host": "pt-4d654cf4576f4d23ad3d3919f12932fe-worker-0",
+  "executable": "/root/local/miniconda3/envs/ragen/bin/python",
+  "cpu_count": 64,
+  "cpu_count_logical": 128,
+  "gpu": "NVIDIA H100 80GB HBM3",
+  "gpu_count": 8,
+  "disk": {
+    "/": {
+      "total": "5153960755200",
+      "used": "447519772672"
+    }
+  },
+  "memory": {
+    "total": "2163642122240"
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-35e2d43d-4067-82ce-90d4-def9e389bf28"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-af4135e3-88f2-e9ac-518d-502c75a85429"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-d7fdeeba-fe9b-ec03-d9f7-6724fe4266b5"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-ccc4f668-3882-5a8e-2c07-c5cd08f6f666"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-7b73c0cf-d3d5-e10c-7176-a43be1e41001"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-81b58d94-5d1f-8ec2-f9d2-fd56172ed177"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-03e8bc66-3b44-6794-49fd-5392fbdda6d1"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-b7cf0ec6-7c29-1179-dceb-09565da51890"
+    }
+  ],
+  "cudaVersion": "12.4",
+  "writerId": "zrvrl6pdc6hzbh0erqwl9hu99oek6ey0"
+}

wandb/run-20251218_182331-3lp68bhy/files/wandb-summary.json

@@ -0,0 +1 @@
+{"train/success_rate":1,"train/policy_loss":0.008788106963038445,"perf/SPS":2270,"_step":101208,"rollout/ep_rew_mean":1,"losses/explained_variance":0.9954127543605864,"train/value_loss":9.734073991296555e-09,"train/learning_rate":0.00022515368852459017,"losses/policy_loss":0.008788106963038445,"charts/SPS":2270,"losses/value_loss":9.734073991296555e-09,"train/old_approx_kl":-0.006179366260766983,"charts/progress":10.040983606557377,"charts/train_success_rate":1,"_timestamp":1.766053461148376e+09,"train/clipfrac":0.03662109375,"_wandb":{"runtime":47},"charts/avg_reward":0.5,"_runtime":47.983079283,"global_step":101208,"train/approx_kl":0.007127484306693077,"rollout/ep_len_mean":1,"rollout/success_rate":1,"charts/avg_value":0.9949655532836914,"train/entropy":0.7567218542098999}

wandb/run-20251218_182448-on6j4c96/files/code/cleanrl/cleanrl/ppo_rubikscube.py

@@ -0,0 +1,517 @@
+# PPO with small MLP for RAGEN Rubik's Cube 2x2 using the existing env (no env edits)
+import os
+import random
+import time
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Tuple, Dict, Any, List
+import json
+import re
+
+import gymnasium as gym
+import numpy as np
+import torch
+import torch.nn as nn
+import torch.optim as optim
+import tyro
+from torch.distributions.categorical import Categorical
+
+import sys
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), '../../'))
+
+from ragen.env.rubikscube.env import RubiksCube2x2Env
+from ragen.env.rubikscube.config import RubiksCube2x2Config
+
+
+class RubiksCubeWrapper(gym.Env):
+    """
+    Adapter to use ragen RubiksCube2x2Env with Gymnasium vector API.
+    - Converts text observation to one-hot vector of 24 stickers x 6 colors.
+    - Maps agent actions [0..11] to env actions [1..12].
+    - Exposes proper observation_space and action_space.
+    """
+    metadata = {"render_modes": ["rgb_array", "human", "ansi"]}
+
+    def __init__(self, env: RubiksCube2x2Env):
+        super().__init__()
+        self._env = env
+        # 24 stickers, 6 colors -> one-hot size 144
+        self._colors = ['W', 'O', 'G', 'R', 'B', 'Y']
+        self._color_to_idx = {c: i for i, c in enumerate(self._colors)}
+        self.observation_space = gym.spaces.Box(low=0.0, high=1.0, shape=(24 * len(self._colors),), dtype=np.float32)
+        self.action_space = gym.spaces.Discrete(12)
+        # precompile regex to extract 4 letters within [X, X]\n [X, X]
+        # Lines look like: "Up (U):    [W, W]\n       [W, W]"
+        self._face_pat = re.compile(r"\[(?:\s*([A-Z])\s*,\s*([A-Z])\s*\])\n\s*\[(?:\s*([A-Z])\s*,\s*([A-Z])\s*\])")
+
+    def _encode_obs(self, text_obs: str) -> np.ndarray:
+        # Extract the six face blocks in the order U, L, F, R, B, D as rendered
+        faces_order = ["Up (U):", "Left (L):", "Front (F):", "Right (R):", "Back (B):", "Down (D):"]
+        onehots: List[int] = []
+        # Build a map from header to its block text
+        lines = text_obs.splitlines()
+        # Collect blocks starting after header line and including the next line for second row
+        i = 0
+        blocks: List[str] = []
+        while i < len(lines):
+            line = lines[i]
+            for header in faces_order:
+                if line.startswith(header):
+                    # Join current line's bracketed pair and the next line (which contains the second pair)
+                    # Remove the "Header:" prefix to keep only the bracket portions
+                    content = line[len(header):].strip()
+                    next_line = lines[i + 1] if i + 1 < len(lines) else ""
+                    block = f"{content}\n{next_line}"
+                    blocks.append(block)
+                    break
+            i += 1
+        # Fallback: if regex fails, return zeros
+        if len(blocks) != 6:
+            return np.zeros(24 * len(self._colors), dtype=np.float32)
+        stickers: List[int] = []
+        for blk in blocks:
+            m = self._face_pat.search(blk)
+            if not m:
+                return np.zeros(24 * len(self._colors), dtype=np.float32)
+            # order: 0,1,2,3 per render() doc
+            c0, c1, c2, c3 = m.group(1), m.group(2), m.group(3), m.group(4)
+            stickers.extend([c0, c1, c2, c3])
+        # stickers now length 24; convert to one-hot
+        grid = np.zeros((24, len(self._colors)), dtype=np.float32)
+        for idx, ch in enumerate(stickers):
+            cidx = self._color_to_idx.get(ch, None)
+            if cidx is not None:
+                grid[idx, cidx] = 1.0
+        return grid.reshape(-1)
+
+    def reset(self, *, seed: int | None = None, options: Dict[str, Any] | None = None):
+        text_obs = self._env.reset(seed=seed)
+        obs = self._encode_obs(text_obs)
+        return obs, {}
+
+    def step(self, action: int):
+        mapped = int(action) + 1  # 0..11 -> 1..12
+        text_obs, reward, done, info = self._env.step(mapped)
+        obs = self._encode_obs(text_obs)
+        terminated = bool(done)
+        truncated = False
+        return obs, float(reward), terminated, truncated, info or {}
+
+    def render(self):
+        return self._env.render()
+
+    def close(self):
+        self._env.close()
+
+
+@dataclass
+class Args:
+    exp_name: str = os.path.basename(__file__)[: -len(".py")]
+    seed: int = 1
+    torch_deterministic: bool = True
+    cuda: bool = True
+    track: bool = True
+    wandb_project_name: str = "cleanRL"
+    wandb_entity: str | None = None
+    capture_video: bool = False
+
+    # Algorithm
+    env_id: str = "RubiksCube2x2"
+    total_timesteps: int = 1000_000
+    learning_rate: float = 2.5e-4
+    num_envs: int = 8
+    num_steps: int = 128
+    anneal_lr: bool = True
+    gamma: float = 0.99
+    gae_lambda: float = 0.95
+    num_minibatches: int = 4
+    update_epochs: int = 4
+    norm_adv: bool = True
+    clip_coef: float = 0.2
+    clip_vloss: bool = True
+    ent_coef: float = 0.01
+    vf_coef: float = 0.5
+    max_grad_norm: float = 0.5
+    target_kl: float | None = None
+
+    # Rubik specific
+    scramble_depth: int = 2
+    max_steps_env: int = 20
+
+    # runtime filled
+    batch_size: int = 0
+    minibatch_size: int = 0
+    num_iterations: int = 0
+    eval_splits: int = 2
+    eval_episodes: int = 4000
+
+
+def make_env(idx, run_name, seed, scramble_depth, max_steps_env, capture_video=False):
+    def thunk():
+        config = RubiksCube2x2Config(scramble_depth=scramble_depth, max_steps=max_steps_env, render_mode='text')
+        env = RubiksCube2x2Env(config)
+        env = RubiksCubeWrapper(env)
+        env = gym.wrappers.TimeLimit(env, max_episode_steps=max_steps_env)
+        env = gym.wrappers.RecordEpisodeStatistics(env)
+        if capture_video and idx == 0:
+            env = gym.wrappers.RecordVideo(env, f"videos/{run_name}")
+        return env
+    return thunk
+
+
+def layer_init(layer, std=np.sqrt(2), bias_const=0.0):
+    torch.nn.init.orthogonal_(layer.weight, std)
+    torch.nn.init.constant_(layer.bias, bias_const)
+    return layer
+
+
+class Agent(nn.Module):
+    def __init__(self, envs):
+        super().__init__()
+        obs_shape = int(np.array(envs.single_observation_space.shape).prod())
+        hidden = 128
+        self.critic = nn.Sequential(
+            layer_init(nn.Linear(obs_shape, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, 1), std=1.0),
+        )
+        self.actor = nn.Sequential(
+            layer_init(nn.Linear(obs_shape, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, envs.single_action_space.n), std=0.01),
+        )
+
+    def get_value(self, x):
+        return self.critic(x)
+
+    def get_action_and_value(self, x, action=None):
+        logits = self.actor(x)
+        probs = Categorical(logits=logits)
+        if action is None:
+            action = probs.sample()
+        return action, probs.log_prob(action), probs.entropy(), self.critic(x)
+
+
+if __name__ == "__main__":
+    args = tyro.cli(Args)
+    args.batch_size = int(args.num_envs * args.num_steps)
+    args.minibatch_size = int(args.batch_size // args.num_minibatches)
+    args.num_iterations = args.total_timesteps // args.batch_size
+    run_name = f"{args.env_id}__{args.exp_name}__{args.seed}__{int(time.time())}"
+
+    if args.track:
+        import wandb
+        wandb.init(
+            project=args.wandb_project_name,
+            entity=args.wandb_entity,
+            config=vars(args),
+            name=run_name,
+            monitor_gym=True,
+            save_code=True,
+        )
+        try:
+            wandb.define_metric("global_step")
+            for prefix in ["train/*", "rollout/*", "eval/*", "losses/*", "charts/*", "perf/*"]:
+                wandb.define_metric(prefix, step_metric="global_step")
+        except Exception:
+            pass
+
+    # seeding
+    random.seed(args.seed)
+    np.random.seed(args.seed)
+    torch.manual_seed(args.seed)
+    torch.backends.cudnn.deterministic = args.torch_deterministic
+
+    device = torch.device("cuda" if torch.cuda.is_available() and args.cuda else "cpu")
+
+    # envs
+    envs = gym.vector.SyncVectorEnv([
+        make_env(i, run_name, args.seed, args.scramble_depth, args.max_steps_env, args.capture_video)
+        for i in range(args.num_envs)
+    ])
+    assert isinstance(envs.single_action_space, gym.spaces.Discrete)
+
+    agent = Agent(envs).to(device)
+    optimizer = optim.Adam(agent.parameters(), lr=args.learning_rate, eps=1e-5)
+
+    # storage
+    obs = torch.zeros((args.num_steps, args.num_envs) + envs.single_observation_space.shape).to(device)
+    actions = torch.zeros((args.num_steps, args.num_envs) + envs.single_action_space.shape).to(device)
+    logprobs = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    rewards = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    dones = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    values = torch.zeros((args.num_steps, args.num_envs)).to(device)
+
+    # start
+    global_step = 0
+    start_time = time.time()
+    next_obs, _ = envs.reset(seed=args.seed)
+    next_obs = torch.Tensor(next_obs).to(device)
+    next_done = torch.zeros(args.num_envs).to(device)
+
+    # eval helper similar to FrozenLake: collect greedy eval trajectories and write metrics.json
+    def collect_eval_trajectories(agent_model, make_env_fn, n_episodes, step_tag):
+        out_dir = Path(f"runs/{run_name}/trajectories/step_{step_tag}")
+        out_dir.mkdir(parents=True, exist_ok=True)
+        out_path = out_dir / "trajectories.jsonl"
+        env = make_env_fn()
+        collected = 0
+        summary_returns = []
+        summary_success = []
+        with out_path.open("w") as f:
+            while collected < n_episodes:
+                state, _ = env.reset(seed=args.seed + 100000 + collected)
+                traj_states = [state.tolist()]
+                traj_actions = []
+                traj_rewards = []
+                traj_dones = []
+                traj_success = []
+                done = False
+                step_count = 0
+                max_eval_steps = getattr(env, '_max_episode_steps', None) or int(args.max_steps_env)
+                while not done:
+                    with torch.no_grad():
+                        logits = agent_model.actor(torch.tensor(state, dtype=torch.float32, device=device).unsqueeze(0))
+                        action = int(torch.argmax(logits, dim=1).item())
+                    next_state, reward, terminated, truncated, info = env.step(action)
+                    traj_actions.append(int(action))
+                    traj_rewards.append(float(reward))
+                    step_count += 1
+                    d = bool(terminated) or bool(truncated) or (step_count >= max_eval_steps)
+                    traj_dones.append(d)
+                    traj_success.append(bool((info or {}).get('success', False)))
+                    state = next_state
+                    traj_states.append(state.tolist())
+                    done = d
+                ep_ret = float(sum(traj_rewards))
+                ep_succ = bool(any(traj_success))
+                record = {
+                    "states": traj_states,
+                    "actions": traj_actions,
+                    "rewards": traj_rewards,
+                    "dones": traj_dones,
+                    "success": traj_success,
+                    "episode_return": ep_ret,
+                    "episode_success": ep_succ,
+                }
+                f.write(json.dumps(record) + "\n")
+                collected += 1
+                summary_returns.append(ep_ret)
+                summary_success.append(1.0 if ep_succ else 0.0)
+        env.close()
+        try:
+            metrics = {
+                "global_step": int(step_tag),
+                "episodes": int(n_episodes),
+                "success_rate": float(np.mean(summary_success)) if len(summary_success) else 0.0,
+                "avg_return": float(np.mean(summary_returns)) if len(summary_returns) else 0.0,
+                "std_return": float(np.std(summary_returns)) if len(summary_returns) else 0.0,
+            }
+            with (out_dir / "metrics.json").open("w") as mf:
+                json.dump(metrics, mf)
+        except Exception as e:
+            print(f"Warning: failed to write eval metrics: {e}")
+
+    # training loop
+    eval_every_iters = max(1, args.num_iterations // args.eval_splits)
+    for iteration in range(1, args.num_iterations + 1):
+        # Anneal LR
+        if args.anneal_lr:
+            frac = 1.0 - (iteration - 1.0) / args.num_iterations
+            lrnow = frac * args.learning_rate
+            optimizer.param_groups[0]["lr"] = lrnow
+
+        # accumulate per-iteration episode successes
+        iter_successes = []
+        for step in range(0, args.num_steps):
+            global_step += args.num_envs
+            obs[step] = next_obs
+            dones[step] = next_done
+
+            with torch.no_grad():
+                action, logprob, _, value = agent.get_action_and_value(next_obs)
+                values[step] = value.flatten()
+            actions[step] = action
+            logprobs[step] = logprob
+
+            next_obs, reward, terminations, truncations, infos = envs.step(action.cpu().numpy())
+            next_done = np.logical_or(terminations, truncations)
+            rewards[step] = torch.tensor(reward).to(device).view(-1)
+            next_obs, next_done = torch.Tensor(next_obs).to(device), torch.Tensor(next_done).to(device)
+
+            # Episode stats logging similar to FrozenLake
+            try:
+                mask = None
+                if isinstance(infos, dict):
+                    if "_episode" in infos:
+                        mask = np.asarray(infos["_episode"]).astype(bool)
+                    elif "episode" in infos and isinstance(infos["episode"], dict) and "_l" in infos["episode"]:
+                        mask = np.asarray(infos["episode"]["_l"]).astype(bool)
+                if mask is not None and np.any(mask):
+                    r_arr = np.asarray(infos.get("episode", {}).get("r", np.zeros_like(mask, dtype=float)))
+                    l_arr = np.asarray(infos.get("episode", {}).get("l", np.zeros_like(mask, dtype=int)))
+                    # prefer success from info; fallback to ep return > 0
+                    if "success" in infos:
+                        succ_arr = np.asarray(infos.get("success", np.zeros_like(mask, dtype=bool))).astype(float)
+                    else:
+                        try:
+                            succ_arr = (np.asarray(r_arr) > 0).astype(float)
+                        except Exception:
+                            succ_arr = np.zeros_like(mask, dtype=float)
+                    # collect iteration successes for training success rate
+                    try:
+                        for s in np.asarray(succ_arr)[mask]:
+                            iter_successes.append(float(s))
+                    except Exception:
+                        pass
+                    if args.track:
+                        try:
+                            import wandb
+                            log_dict = {
+                                "global_step": int(global_step),
+                                "rollout/ep_rew_mean": float(np.mean(r_arr[mask])) if np.any(mask) else None,
+                                "rollout/ep_len_mean": float(np.mean(l_arr[mask])) if np.any(mask) else None,
+                                "rollout/success_rate": float(np.mean(succ_arr[mask])) if np.any(mask) else None,
+                            }
+                            wandb.log(log_dict, step=global_step)
+                        except Exception:
+                            pass
+            except Exception:
+                pass
+
+        # GAE
+        with torch.no_grad():
+            next_value = agent.get_value(next_obs).reshape(1, -1)
+            advantages = torch.zeros_like(rewards).to(device)
+            lastgaelam = 0
+            for t in reversed(range(args.num_steps)):
+                if t == args.num_steps - 1:
+                    nextnonterminal = 1.0 - next_done
+                    nextvalues = next_value
+                else:
+                    nextnonterminal = 1.0 - dones[t + 1]
+                    nextvalues = values[t + 1]
+                delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
+                advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
+            returns = advantages + values
+
+        # flatten batch
+        b_obs = obs.reshape((-1,) + envs.single_observation_space.shape)
+        b_logprobs = logprobs.reshape(-1)
+        b_actions = actions.reshape((-1,) + envs.single_action_space.shape)
+        b_advantages = advantages.reshape(-1)
+        b_returns = returns.reshape(-1)
+        b_values = values.reshape(-1)
+
+        # update
+        b_inds = np.arange(args.batch_size)
+        clipfracs = []
+        for epoch in range(args.update_epochs):
+            np.random.shuffle(b_inds)
+            for start in range(0, args.batch_size, args.minibatch_size):
+                end = start + args.minibatch_size
+                mb_inds = b_inds[start:end]
+
+                _, newlogprob, entropy, newvalue = agent.get_action_and_value(b_obs[mb_inds], b_actions.long()[mb_inds])
+                logratio = newlogprob - b_logprobs[mb_inds]
+                ratio = logratio.exp()
+
+                with torch.no_grad():
+                    old_approx_kl = (-logratio).mean()
+                    approx_kl = ((ratio - 1) - logratio).mean()
+                    clipfracs += [((ratio - 1.0).abs() > args.clip_coef).float().mean().item()]
+
+                mb_advantages = b_advantages[mb_inds]
+                if args.norm_adv:
+                    mb_advantages = (mb_advantages - mb_advantages.mean()) / (mb_advantages.std() + 1e-8)
+
+                pg_loss1 = -mb_advantages * ratio
+                pg_loss2 = -mb_advantages * torch.clamp(ratio, 1 - args.clip_coef, 1 + args.clip_coef)
+                pg_loss = torch.max(pg_loss1, pg_loss2).mean()
+
+                newvalue = newvalue.view(-1)
+                if args.clip_vloss:
+                    v_loss_unclipped = (newvalue - b_returns[mb_inds]) ** 2
+                    v_clipped = b_values[mb_inds] + torch.clamp(
+                        newvalue - b_values[mb_inds], -args.clip_coef, args.clip_coef,
+                    )
+                    v_loss_clipped = (v_clipped - b_returns[mb_inds]) ** 2
+                    v_loss = 0.5 * torch.max(v_loss_unclipped, v_loss_clipped).mean()
+                else:
+                    v_loss = 0.5 * ((newvalue - b_returns[mb_inds]) ** 2).mean()
+
+                entropy_loss = entropy.mean()
+                loss = pg_loss - args.ent_coef * entropy_loss + v_loss * args.vf_coef
+
+                optimizer.zero_grad()
+                loss.backward()
+                nn.utils.clip_grad_norm_(agent.parameters(), args.max_grad_norm)
+                optimizer.step()
+
+            if args.target_kl is not None and approx_kl > args.target_kl:
+                break
+
+        # metrics
+        y_pred, y_true = b_values.cpu().numpy(), b_returns.cpu().numpy()
+        var_y = np.var(y_true)
+        explained_var = np.nan if var_y == 0 else 1 - np.var(y_true - y_pred) / var_y
+
+        sps = int(global_step / (time.time() - start_time))
+        train_success_rate = float(np.mean(iter_successes)) if len(iter_successes) else 0.0
+        print(f"Iter {iteration:4d}/{args.num_iterations} | SPS: {sps:5d} | R: {rewards.mean().item():6.3f}")
+        if args.track:
+            try:
+                import wandb
+                wandb.log({
+                    "global_step": int(global_step),
+                    "charts/progress": float(100.0 * iteration / max(1, args.num_iterations)),
+                    "train/value_loss": float(v_loss.item()),
+                    "train/policy_loss": float(pg_loss.item()),
+                    "losses/value_loss": float(v_loss.item()),
+                    "losses/policy_loss": float(pg_loss.item()),
+                    "train/entropy": float(entropy_loss.item()),
+                    "train/old_approx_kl": float(old_approx_kl.item()),
+                    "train/approx_kl": float(approx_kl.item()),
+                    "train/clipfrac": float(np.mean(clipfracs)) if len(clipfracs) else 0.0,
+                    "losses/explained_variance": float(explained_var),
+                    "charts/avg_reward": float(rewards.mean().item()),
+                    "charts/avg_value": float(values.mean().item()),
+                    "perf/SPS": int(sps),
+                    "charts/SPS": int(sps),
+                    "train/success_rate": train_success_rate,
+                    "charts/train_success_rate": train_success_rate,
+                    "train/learning_rate": float(optimizer.param_groups[0]["lr"]),
+                }, step=global_step)
+            except Exception:
+                pass
+
+        # periodic evaluation collection
+        if iteration % eval_every_iters == 0:
+            try:
+                eval_thunk = make_env(0, run_name, args.seed + 9999, args.scramble_depth, args.max_steps_env, False)
+                collect_eval_trajectories(agent, eval_thunk, n_episodes=args.eval_episodes, step_tag=global_step)
+                if args.track:
+                    try:
+                        import json as _json
+                        from pathlib import Path as _Path
+                        mpath = _Path(f"runs/{run_name}/trajectories/step_{global_step}/metrics.json")
+                        if mpath.exists():
+                            with mpath.open("r") as mf:
+                                metrics = _json.load(mf)
+                            wandb.log({
+                                "eval/success_rate": metrics.get("success_rate"),
+                                "eval/avg_return": metrics.get("avg_return"),
+                                "eval/std_return": metrics.get("std_return"),
+                                "eval/episodes": metrics.get("episodes"),
+                            }, step=global_step)
+                    except Exception:
+                        pass
+                print(f"Collected {args.eval_episodes} eval trajectories at global_step {global_step}")
+            except Exception as e:
+                print(f"Warning: eval trajectory collection failed at step {global_step}: {e}")
+
+    envs.close()

wandb/run-20251218_182448-on6j4c96/files/config.yaml

@@ -0,0 +1,201 @@
+_wandb:
+    value:
+        cli_version: 0.23.0
+        code_path: code/cleanrl/cleanrl/ppo_rubikscube.py
+        e:
+            ubi6kfhhs7rtnrt9cd5f40pg33djhcqy:
+                args:
+                    - --track
+                codePath: cleanrl/cleanrl/ppo_rubikscube.py
+                codePathLocal: cleanrl/cleanrl/ppo_rubikscube.py
+                cpu_count: 64
+                cpu_count_logical: 128
+                cudaVersion: "12.4"
+                disk:
+                    /:
+                        total: "5153960755200"
+                        used: "447519776768"
+                email: haoyu-wa22@mails.tsinghua.edu.cn
+                executable: /root/local/miniconda3/envs/ragen/bin/python
+                git:
+                    commit: 8d73639b99b38265453f898b8d6af7d4af50d56e
+                    remote: https://github.com/mll-lab-nu/RAGEN.git
+                gpu: NVIDIA H100 80GB HBM3
+                gpu_count: 8
+                gpu_nvidia:
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-35e2d43d-4067-82ce-90d4-def9e389bf28
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-af4135e3-88f2-e9ac-518d-502c75a85429
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-d7fdeeba-fe9b-ec03-d9f7-6724fe4266b5
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-ccc4f668-3882-5a8e-2c07-c5cd08f6f666
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-7b73c0cf-d3d5-e10c-7176-a43be1e41001
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-81b58d94-5d1f-8ec2-f9d2-fd56172ed177
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-03e8bc66-3b44-6794-49fd-5392fbdda6d1
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-b7cf0ec6-7c29-1179-dceb-09565da51890
+                host: pt-4d654cf4576f4d23ad3d3919f12932fe-worker-0
+                memory:
+                    total: "2163642122240"
+                os: Linux-5.14.0-284.25.1.el9_2.x86_64-x86_64-with-glibc2.35
+                program: /mnt/general/wanghy/RAGEN/cleanrl/cleanrl/ppo_rubikscube.py
+                python: CPython 3.10.19
+                root: /mnt/general/wanghy/RAGEN
+                startedAt: "2025-12-18T10:24:48.379419Z"
+                writerId: ubi6kfhhs7rtnrt9cd5f40pg33djhcqy
+        m:
+            - "1": global_step
+              "6":
+                - 3
+              "7": []
+            - "2": charts/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": perf/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": train/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": rollout/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": eval/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": losses/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+        python_version: 3.10.19
+        t:
+            "1":
+                - 1
+                - 11
+                - 30
+                - 49
+                - 50
+                - 51
+                - 105
+            "2":
+                - 1
+                - 11
+                - 30
+                - 49
+                - 50
+                - 51
+                - 105
+            "3":
+                - 7
+                - 13
+                - 16
+                - 61
+            "4": 3.10.19
+            "5": 0.23.0
+            "6": 4.57.1
+            "12": 0.23.0
+            "13": linux-x86_64
+anneal_lr:
+    value: true
+batch_size:
+    value: 1024
+capture_video:
+    value: false
+clip_coef:
+    value: 0.2
+clip_vloss:
+    value: true
+cuda:
+    value: true
+ent_coef:
+    value: 0.01
+env_id:
+    value: RubiksCube2x2
+eval_episodes:
+    value: 4000
+eval_splits:
+    value: 2
+exp_name:
+    value: ppo_rubikscube
+gae_lambda:
+    value: 0.95
+gamma:
+    value: 0.99
+learning_rate:
+    value: 0.00025
+max_grad_norm:
+    value: 0.5
+max_steps_env:
+    value: 20
+minibatch_size:
+    value: 256
+norm_adv:
+    value: true
+num_envs:
+    value: 8
+num_iterations:
+    value: 976
+num_minibatches:
+    value: 4
+num_steps:
+    value: 128
+scramble_depth:
+    value: 2
+seed:
+    value: 1
+target_kl:
+    value: null
+torch_deterministic:
+    value: true
+total_timesteps:
+    value: 1000000
+track:
+    value: true
+update_epochs:
+    value: 4
+vf_coef:
+    value: 0.5
+wandb_entity:
+    value: null
+wandb_project_name:
+    value: cleanRL

wandb/run-20251218_182448-on6j4c96/files/diff.patch

@@ -0,0 +1,1162 @@
+diff --git a/config/base.yaml b/config/base.yaml
+index a1659e4..9df6250 100644
+--- a/config/base.yaml
++++ b/config/base.yaml
+@@ -9,15 +9,16 @@ seed:
+   train: 10000
+   val: 123
+ 
+-micro_batch_size_per_gpu: 4
++micro_batch_size_per_gpu: 1
+ ppo_mini_batch_size: 32
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/Qwen/Qwen2.5-0.5B-Instruct
++# Qwen/Qwen2.5-0.5B-Instruct
+ enable_response_mask: True # Enabling response mask could improve stability of rollout/old_log_prob, as P(st|history) are no longer calculated in loss here. See https://docs.google.com/document/d/1bg7obeiKTExuHHBl5uOiSpec5uLDZ2Tgvxy6li5pHX4/edit?usp=sharing for more details.
+ grpo_advantage_length_weight: False # if you do not enable this and critic/advantage_estimator is GRPO, and the critic/advantages/mean is too low, then you can try enabling this to encourage reasoning and forbid collapse
+ 
+ lora:
+   rank: 0
+-  alpha: 64
++  alpha: 16
+   target_modules: all-linear
+ 
+ actor_rollout_ref:
+@@ -46,10 +47,10 @@ actor_rollout_ref:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu} # following micro_batch_size_per_gpu
+     tensor_model_parallel_size: 1
+-    max_model_len: 3600
++    max_model_len: 7200 #3600 why** 14400
+     prompt_length: 1 # useless. Just put it here
+     response_length: 400 # single-turn response length
+-    gpu_memory_utilization: 0.5
++    gpu_memory_utilization: 0.7
+     max_num_batched_tokens: 8192 # set only when enable_chunked_prefill is true
+     temperature: 1
+     rollout_filter_ratio: 0.25
+@@ -90,26 +91,28 @@ algorithm:
+     kl_coef: 0.000
+ 
+ trainer:
+-  project_name: ragen_latest
++  project_name: 
+   experiment_name: test
+   local_log_dir: "results/"
+-  save_freq: 100
++  save_freq: -1
+   total_training_steps: 200
+   validation_steps: 1 # validation instances = validation_steps * val_env_groups * group_size
+   val_before_train: True
+   n_gpus_per_node: 1
+   test_freq: 10
+   generations_to_log_to_wandb: 
++    train: 128
+     val: 20
+   logger: [ 'console', 'wandb' ]
+   max_actor_ckpt_to_keep: 1
+   max_critic_ckpt_to_keep: 1
++  default_local_dir: /mnt/general/wanghy/RAGEN/saves/
+ 
+ agent_proxy:
+   max_context_window: -1 # set a value > 0 to enable context window for long trajectory
+-  max_turn: 5
++  max_turn: 25 #25 why** 700
+   action_sep: "||"
+-  max_actions_per_turn: 2 # how many actions can be output at most in a single turn
++  max_actions_per_turn: 1 # how many actions can be output at most in a single turn
+   use_turn_scores: False # important to GAE when applying token-level rewards to token-level advantages. If False, will take the sum of scores as the reward for the last turn.
+   enable_think: True # False -> no think RL
+   reward_normalization:
+@@ -126,11 +129,11 @@ es_manager:
+       tags: ["CoordSokoban"]
+       n_groups: [8] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+   val:
+-    env_groups: 32
+-    group_size: 16 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
++    env_groups: 256
++    group_size: 1 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
+     env_configs:
+       tags: ["CoordSokoban"]
+-      n_groups: [32] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      n_groups: [256] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ ctx_manager:
+   generation: # go to vllm
+diff --git a/config/envs.yaml b/config/envs.yaml
+index 1002ac8..ba3ac6f 100644
+--- a/config/envs.yaml
++++ b/config/envs.yaml
+@@ -2,6 +2,7 @@ custom_envs:
+   SimpleSokoban:
+     env_type: sokoban
+     max_actions_per_traj: 10 # used in environment state manager to control the actual max actions executed per trajectory
++    # ORIGNAL env_instruction
+     env_instruction: |
+       You are solving the Sokoban puzzle. 
+       You are the player and you need to push all boxes to targets. 
+@@ -12,7 +13,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+ 
+   LargerSokoban:
+@@ -65,7 +66,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+       observation_format: "grid_coord"
+ 
+@@ -111,14 +112,14 @@ custom_envs:
+ 
+   FrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: "You are solving the FrozenLake puzzle. Forbid the whole and go to the target. You may move to the unintended direction due to the slippery ice. Example answer format: <think>To forbid the hole and go to the target, I should go left then go up.</think><answer>Left || Up</answer>"
+     max_tokens: 100
+     env_config: null
+ 
+   CoordFrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: |
+       You are solving the FrozenLake puzzle. The observation includes both a symbol grid and zero-indexed coordinates for the start, goal, player, and any holes.
+       Coordinates range from the top-left corner (0, 0) to the bottom-right corner (5, 5).
+@@ -195,4 +196,80 @@ custom_envs:
+     max_actions_per_traj: 30
+     env_instruction: "You are a Lean theorem prover. Given a Lean theorem statement, propose a sequence of tactics that completes the proof. Think step by step about which tactics to apply next. Provide tactics separated by '||', for example <answer>intro || simp || rfl</answer>."
+     max_tokens: 512
+-    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
+
++    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
++
++  
++  game_2048:
++    env_type: game_2048
++    max_actions_per_traj: 700
++    env_instruction: |
++      You are playing the 2048 game on a 4x4 grid. Merge equal tiles by sliding Up, Right, Down, or Left.
++      If a move is invalid (no tiles move), a small penalty is applied. Respond with a single action.
++      Example: <answer>Up</answer>
++    max_tokens: 8192
++    env_config: null
++
++  blackjack:
++    env_type: blackjack
++    max_actions_per_traj: 10
++    env_instruction: |
++      You are playing Blackjack against a dealer. The dealer must hit on 16 or less and stand on 17 or more.
++      Choose either Stick or Hit. Respond with a single action.
++      Example: <answer>Hit</answer>
++    max_tokens: 64
++    env_config: null
++
++  rubikscube:
++    env_type: rubikscube
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.
++      Available actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.
++      - Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).
++      - Modifiers: A letter alone means 90° clockwise (e.g., 'R'). A letter with prime (') means 90° counter-clockwise (e.g., "R'").
++      Respond with a sequence of actions separated by "||".
++      Example: <answer>U</answer>
++    max_tokens: 96
++    env_config:
++      scramble_depth: 1
++      max_steps: 20
++      render_mode: "text"
++  
++  SimpleSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 2x2 box contains the numbers 1-4 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format, for example: <answer>place 1 at row 2 col 3</answer> or <answer>1,2,3</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 4
++      difficulty: "easy"
++      render_format: "with_feedback"
++      show_conflicts: false
++      show_valid_numbers: false
++      max_steps: 20
++
++  MediumSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 30
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 3x3 box contains the numbers 1-9 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format: <answer>place 5 at row 2 col 3</answer> or <answer>2,3,5</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 9
++      difficulty: "medium"
++      render_format: "with_feedback"
++      show_conflicts: true
++      show_valid_numbers: true
++      max_steps: 81
++
+diff --git a/config/eval.yaml b/config/eval.yaml
+index e65f3a5..7775fd1 100644
+--- a/config/eval.yaml
++++ b/config/eval.yaml
+@@ -1,14 +1,15 @@
+ defaults:
+   - envs
+-
++enable_response_mask: True
+ system:
+-  CUDA_VISIBLE_DEVICES: "0"
++  CUDA_VISIBLE_DEVICES: "0,1,2,3,4,5,6,7"
+ 
+ seed:
+   train: 10000
+   val: 123
+ 
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/openai/gpt-oss-20b
++# /mnt/general/share/model/tyzhu/SPA-frozenlake-qwen2.5-1.5b-instruct
+ 
+ lora:
+   rank: 0
+@@ -24,9 +25,9 @@ actor_rollout_ref:
+   rollout:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu}
+-    tensor_model_parallel_size: 1
++    tensor_model_parallel_size: 8
+     dtype: bfloat16
+-    max_model_len: 3600
++    max_model_len: 7200
+     prompt_length: 1
+     response_length: 400
+     gpu_memory_utilization: 0.9
+@@ -35,6 +36,7 @@ actor_rollout_ref:
+     free_cache_engine: True
+     enable_chunked_prefill: False
+     disable_log_stats: False
++    do_sample: True
+     val_kwargs:
+       do_sample: True
+       temperature: 0.5
+@@ -44,9 +46,9 @@ actor_rollout_ref:
+ 
+ agent_proxy:
+   max_context_window: -1
+-  max_turn: 5
++  max_turn: 25
+   action_sep: "||"
+-  max_actions_per_turn: 2
++  max_actions_per_turn: 1
+   use_turn_scores: False
+   enable_think: True
+   reward_normalization:
+@@ -59,13 +61,13 @@ es_manager:
+     env_groups: 8
+     group_size: 16
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [8]
+   val:
+     env_groups: 32
+-    group_size: 16
++    group_size: 128
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [32]
+ 
+ ctx_manager:
+diff --git a/config/evaluate_api_llm.yaml b/config/evaluate_api_llm.yaml
+index b21c95e..e2e6761 100644
+--- a/config/evaluate_api_llm.yaml
++++ b/config/evaluate_api_llm.yaml
+@@ -1,8 +1,11 @@
++#export OPENAI_BASE_URL="https://api.ohmygpt.com/v1"
++#export OPENAI_API_KEY="sk-o4sMxBkN5BB100C4D4a3T3BlBkFJF7791CA39EA14ca98041"
++#python -m ragen.eval_api hydra.searchpath='[file://./verl/verl/trainer/config]'
+ defaults:
+   - base # this is a symbolic link to the verl/verl/trainer/config/ppo_trainer.yaml file
+ 
+ model_config:
+-  model_name: gpt-4o # should be registered in model_info
++  model_name: ark-deepseek-v3-250324 # should be registered in model_info
+   max_concurrency: 16
+ 
+ model_info:
+@@ -24,9 +27,9 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+-  gpt-4o:
++  gpt-4o-mini:
+     provider_name: openai
+-    model_name: gpt-4o
++    model_name: gpt-4o-mini
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+@@ -36,21 +39,42 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  ark-deepseek-v3-250324:
++    provider_name: openai
++    model_name: ark-deepseek-v3-250324
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
+   deepseek-v3:
+     provider_name: deepseek
+     model_name: deepseek-chat
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  glm-4.6:
++    provider_name: openai
++    model_name: glm-4.6
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
++  TA/openai/gpt-oss-120b:
++    provider_name: openai
++    model_name: TA/openai/gpt-oss-120b
++    generation_kwargs:
++      temperature: 0
++      max_tokens: 8192
++      # max_retries: 5
+ 
+-
+-
++agent_proxy:
++  max_turn: 5
+ es_manager:
+   val:
+-    env_groups: 256
++    env_groups: 128
+     group_size: 1 # should be set to 1 because val temperature is set to 0 and same prompt leads to same output
+     env_configs:
+-      tags: ["CoordSokoban"]
+-      n_groups: [256] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      tags: ["rubikscube"]
++      n_groups: [128] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ 
++rollout:
++  max_model_len: 7200
+
+Submodule external/webshop-minimal contains modified content
+diff --git a/external/webshop-minimal/requirements.txt b/external/webshop-minimal/requirements.txt
+index 5a1b04f..238ed5a 100644
+--- a/external/webshop-minimal/requirements.txt
++++ b/external/webshop-minimal/requirements.txt
+@@ -4,7 +4,7 @@ flask
+ html2text
+ rank_bm25
+ pyserini
+-faiss-cpu
++faiss-gpu
+ thefuzz
+ gdown
+ spacy
+diff --git a/ragen/env/__init__.py b/ragen/env/__init__.py
+index b0f3461..bc6838e 100644
+--- a/ragen/env/__init__.py
++++ b/ragen/env/__init__.py
+@@ -12,6 +12,14 @@ from .metamathqa.env import MetaMathQAEnv
+ from .metamathqa.config import MetaMathQAEnvConfig
+ from .lean.config import LeanEnvConfig
+ from .lean.env import LeanEnv
++from .game_2048.config import Game2048EnvConfig
++from .game_2048.env import Game2048Env
++from .blackjack.config import BlackjackEnvConfig
++from .blackjack.env import BlackjackEnv
++from .rubikscube.config import RubiksCube2x2Config
++from .rubikscube.env import RubiksCube2x2Env
++from .sudoku.config import SudokuEnvConfig
++from .sudoku.env import SudokuEnv
+ 
+ 
+ REGISTERED_ENVS = {
+@@ -22,6 +30,10 @@ REGISTERED_ENVS = {
+     # 'alfworld': AlfredTXTEnv,
+     'metamathqa': MetaMathQAEnv,
+     'lean': LeanEnv,
++    'game_2048': Game2048Env,
++    'blackjack': BlackjackEnv,
++    'rubikscube': RubiksCube2x2Env,
++    'sudoku': SudokuEnv,
+ }
+ 
+ REGISTERED_ENV_CONFIGS = {
+@@ -32,6 +44,10 @@ REGISTERED_ENV_CONFIGS = {
+     # 'alfworld': AlfredEnvConfig,
+     'metamathqa': MetaMathQAEnvConfig,
+     'lean': LeanEnvConfig,
++    'game_2048': Game2048EnvConfig,
++    'blackjack': BlackjackEnvConfig,
++    'rubikscube': RubiksCube2x2Config,
++    'sudoku': SudokuEnvConfig,
+ }
+ 
+ try:
+diff --git a/ragen/env/frozen_lake/config.py b/ragen/env/frozen_lake/config.py
+index de054f4..9950c34 100644
+--- a/ragen/env/frozen_lake/config.py
++++ b/ragen/env/frozen_lake/config.py
+@@ -8,7 +8,7 @@ class FrozenLakeEnvConfig:
+     size: int = 4
+     p: float = 0.9
+     success_rate: float = 0.8
+-    is_slippery: bool = True
++    is_slippery: bool = False
+     map_seed: Optional[int] = None
+     render_mode: str = "text"
+     observation_format: str = "grid"
+diff --git a/ragen/env/frozen_lake/env.py b/ragen/env/frozen_lake/env.py
+index 9ee3add..eef11d8 100644
+--- a/ragen/env/frozen_lake/env.py
++++ b/ragen/env/frozen_lake/env.py
+@@ -13,6 +13,7 @@ from ragen.env.base import BaseDiscreteActionEnv
+ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+     def __init__(self, config: FrozenLakeEnvConfig = FrozenLakeEnvConfig()):
+         # Using mappings directly from config
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.GRID_LOOKUP = config.grid_lookup
+         self.ACTION_LOOKUP = config.action_lookup
+@@ -95,6 +96,7 @@ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+ if __name__ == "__main__":
+     import matplotlib.pyplot as plt
+     config = FrozenLakeEnvConfig(size=4, is_slippery=True)
++    # import pdb;pdb.set_trace()
+     env = FrozenLakeEnv(config)
+     print(env.reset())
+     while True:
+diff --git a/ragen/env/sokoban/env.py b/ragen/env/sokoban/env.py
+index 17cd636..6922b54 100644
+--- a/ragen/env/sokoban/env.py
++++ b/ragen/env/sokoban/env.py
+@@ -33,6 +33,7 @@ class SokobanEnv(BaseDiscreteActionEnv, GymSokobanEnv):
+     def reset(self, seed=None, mode=None):
+         try:
+             with all_seed(seed):
++                # import pdb;pdb.set_trace()
+                 self.room_fixed, self.room_state, self.box_mapping, action_sequence = generate_room(
+                     dim=self.dim_room,
+                     num_steps=self.num_gen_steps,
+diff --git a/ragen/llm_agent/agent_proxy.py b/ragen/llm_agent/agent_proxy.py
+index c25b50d..f6c4088 100644
+--- a/ragen/llm_agent/agent_proxy.py
++++ b/ragen/llm_agent/agent_proxy.py
+@@ -287,6 +287,7 @@ def main(config):
+ 	proxy = LLMAgentProxy(config, actor_wg, tokenizer)
+ 	import time
+ 	start_time = time.time()
++	# import pdb;pdb.set_trace()
+ 	rollouts = proxy.rollout(
+ 		DataProto(
+ 			batch=None,
+diff --git a/ragen/llm_agent/base_llm.py b/ragen/llm_agent/base_llm.py
+index 358a2eb..b6f7797 100644
+--- a/ragen/llm_agent/base_llm.py
++++ b/ragen/llm_agent/base_llm.py
+@@ -33,6 +33,7 @@ class OpenAIProvider(LLMProvider):
+             raise ValueError("OpenAI API key not provided and not found in environment variables")
+         
+         self.client = AsyncOpenAI(api_key=self.api_key)
++        # import pdb;pdb.set_trace()
+     
+     async def generate(self, messages: List[Dict[str, str]], **kwargs) -> LLMResponse:
+         if "o1-mini" in self.model_name:
+@@ -46,10 +47,7 @@ class OpenAIProvider(LLMProvider):
+         )
+         if response.choices[0].finish_reason in ['length', 'content_filter']:
+             raise ValueError("Content filtered or length exceeded")
+-        return LLMResponse(
+-            content=response.choices[0].message.content,
+-            model_name=response.model
+-        )
++        return LLMResponse(content=response.choices[0].message.content,model_name=response.model)
+ 
+ class DeepSeekProvider(LLMProvider):
+     """DeepSeek API provider implementation"""
+@@ -199,7 +197,7 @@ class ConcurrentLLM:
+         
+         # Queue to store unfinished or failed tasks
+         current_batch = messages_list.copy()
+-        max_retries = kwargs.get("max_retries", 100)
++        max_retries = kwargs.get("max_retries", 10)
+         retry_count = 0
+         
+         while current_batch and retry_count < max_retries:
+diff --git a/ragen/llm_agent/ctx_manager.py b/ragen/llm_agent/ctx_manager.py
+index 905247a..af20a01 100644
+--- a/ragen/llm_agent/ctx_manager.py
++++ b/ragen/llm_agent/ctx_manager.py
+@@ -88,6 +88,7 @@ class ContextManager:
+         Initialize the ContextManager.
+         Processor is used to process the image data.
+         """
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.tokenizer = tokenizer
+         self.processor = processor
+@@ -311,7 +312,7 @@ class ContextManager:
+ 
+             llm_input_texts.append(text_with_prompt)
+             messages_list.append(messages)
+-
++        # import pdb;pdb.set_trace()
+         inputs = self.tokenizer(llm_input_texts, return_tensors="pt", padding=True, padding_side="left", truncation=False) # We have truncated previously, truncation in tokenizer may cause issues.
+         input_ids, attention_mask = inputs.input_ids, inputs.attention_mask
+         position_ids = (attention_mask.cumsum(dim=-1) - 1).clamp(min=0)
+@@ -353,6 +354,30 @@ class ContextManager:
+                 key: np.sum(value) / self.env_nums[key.split("/")[0]]
+                 for key, value in metrics.items()
+             }
++            # Derived metrics for wandb logging
++            try:
++                # charts/avg_episode_return: average across all 2048 env groups
++                ep_keys = [k for k in metrics.keys() if k.endswith('/episodic_return')]
++                if len(ep_keys) > 0:
++                    ep_vals = []
++                    for k in ep_keys:
++                        tag = k.split('/')[0]
++                        denom = self.env_nums.get(tag, max(1, len(metrics[k])))
++                        ep_vals.append(float(np.sum(metrics[k]) / denom))
++                    mean_metrics["charts/avg_episode_return"] = float(np.mean(ep_vals))
++            except Exception:
++                pass
++            try:
++                # rollout/max_tile: max over all envs in this batch
++                tile_keys = [k for k in metrics.keys() if k.endswith('/max_tile')]
++                if len(tile_keys) > 0:
++                    tile_vals = []
++                    for k in tile_keys:
++                        tile_vals.extend(metrics[k])
++                    if len(tile_vals) > 0:
++                        mean_metrics["rollout/max_tile"] = int(np.max(tile_vals))
++            except Exception:
++                pass
+             for key, values in metrics.items():
+                 if not isinstance(values, list):
+                     continue
+diff --git a/ragen/llm_agent/es_manager.py b/ragen/llm_agent/es_manager.py
+index b87a2b3..1a49d3e 100644
+--- a/ragen/llm_agent/es_manager.py
++++ b/ragen/llm_agent/es_manager.py
+@@ -128,18 +128,40 @@ class EnvStateManager:
+         env_outputs: List[Dict]
+             {env_id: int, history: List[Dict][{state: str, actions: List[str], reward: float, info: Dict, llm_response: str, llm_raw_response: str, (Optional)images: List[PIL.Image.Image]}]}
+         """
++        # def _execute_actions(env, actions):
++        #     acc_reward, turn_info, turn_done = 0, {}, False
++        #     executed_actions = []
++        #     for action in actions:
++        #         _, reward, done, info = env.step(action)
++        #         acc_reward += reward
++        #         turn_info.update(info) # NOTE: currently use last info for multi-action
++        #         executed_actions.append(action)
++        #         if done:
++        #             turn_done = True
++        #             break
++            
++        #     return acc_reward, turn_info, turn_done, executed_actions
+         def _execute_actions(env, actions):
+-            acc_reward, turn_info, turn_done = 0, {}, False
++            acc_reward, turn_info, turn_done = 0.0, {}, False
++            raw_acc_reward = 0.0
+             executed_actions = []
+             for action in actions:
+                 _, reward, done, info = env.step(action)
+-                acc_reward += reward
++                acc_reward += float(reward)
++                try:
++                    raw_acc_reward += float(info.get('raw_reward', 0.0))
++                except Exception:
++                    pass
+                 turn_info.update(info) # NOTE: currently use last info for multi-action
+                 executed_actions.append(action)
+                 if done:
+                     turn_done = True
+                     break
+-            
++            # Overwrite per-turn raw_reward to reflect the sum across all executed actions in this turn
++            try:
++                turn_info['raw_reward'] = float(raw_acc_reward)
++            except Exception:
++                pass
+             return acc_reward, turn_info, turn_done, executed_actions
+ 
+         def _log_env_state(status, history, cur_obs, max_actions_per_traj, executed_actions, all_actions, acc_reward, turn_done, turn_info, env_input):
+@@ -198,6 +220,18 @@ class EnvStateManager:
+                 'success': float(status.terminated and (not status.truncated)),
+                 'num_actions': status.num_actions,
+             }
++            # Add episodic-level metrics
++            # try:
++            #     # Sum of per-turn rewards equals the episodic return (env-shaped reward)
++            #     env_metric['episodic_return'] = float(sum(status.rewards))
++            # except Exception:
++            #     pass
++            try:
++                # Final max tile on the board at the end of the rollout
++                import numpy as _np
++                env_metric['max_tile'] = int(_np.max(entry['env'].grid))
++            except Exception:
++                pass
+             custom_metric = {}
+             for turn in cache['history']:
+                 for k, v in turn.get('info', {}).items():
+@@ -212,6 +246,12 @@ class EnvStateManager:
+                             "Skipping non-numeric metric '%s' with value %r for env %s.",
+                             k, v, entry['tag']
+                         )
++            # Add episodic_return as the SUM of raw_reward across steps (align with CleanRL)
++            try:
++                if 'raw_reward' in custom_metric:
++                    env_metric['episodic_return'] = float(np.sum(custom_metric['raw_reward']))
++            except Exception:
++                pass
+             for k, v in custom_metric.items():
+                 # TODO: Move TURN_LVL_METRICS into the environment
+                 if "webshop" not in cache['tag'].lower() or ("webshop" in cache['tag'].lower() and k in TURN_LVL_METRICS):
+@@ -219,7 +259,12 @@ class EnvStateManager:
+                 else:
+                     env_metric['traj_sum/' + k] = np.sum(v)
+ 
+-
++            try:
++                if 'score' in custom_metric and len(custom_metric['score']) > 0:
++                    env_metric['final_score'] = float(custom_metric['score'][-1])
++            except Exception:
++                pass
++                
+             cache['history'][-1]['metrics'] = custom_metric
+             env_metric = {f"{entry['tag']}/{k}": v for k, v in env_metric.items()}
+             cache['metrics'] = env_metric
+diff --git a/requirements.txt b/requirements.txt
+index 2fd756e..9b521bf 100644
+--- a/requirements.txt
++++ b/requirements.txt
+@@ -7,7 +7,6 @@ accelerate
+ codetiming
+ datasets
+ dill
+-flash-attn==2.7.4.post1
+ hydra-core
+ numpy
+ pandas
+@@ -15,19 +14,19 @@ pybind11
+ ray>=2.10
+ tensordict>=0.8.0,<0.9.0
+ transformers
+-vllm==0.8.2
++vllm==0.8.5
+ wandb
+ gymnasium
+ gymnasium[toy-text]
+ 
+ pyarrow>=15.0.0
+ pylatexenc
+-torchdata
++# torchdata
+ debugpy
+ 
+ together
+ anthropic
+-faiss-cpu==1.11.0
++faiss-gpu
+ 
+ # This is optional, but needs to be installed with main requirements if you want to use webshop
+ -r external/webshop-minimal/requirements.txt
+diff --git a/scripts/setup_ragen.sh b/scripts/setup_ragen.sh
+index f9a7cd9..85a93c4 100644
+--- a/scripts/setup_ragen.sh
++++ b/scripts/setup_ragen.sh
+@@ -94,10 +94,10 @@ main() {
+         pip install torch==2.5.0 --index-url https://download.pytorch.org/whl/cu124
+         
+         print_step "Installing flash-attention..."
+-        pip3 install flash-attn==2.7.4.post1 --no-build-isolation
++        # pip3 install flash-attn==2.7.4.post1 --no-build-isolation
+     else
+         print_step "Installing PyTorch without CUDA support..."
+-        pip install torch==2.5.0
++        pip install torch==2.4.0
+     fi
+     
+     # Install remaining requirements
+@@ -137,8 +137,8 @@ main() {
+     conda install conda-forge::gdown
+     mkdir -p external/webshop-minimal/webshop_minimal/data/full
+     cd external/webshop-minimal/webshop_minimal/data/full
+-    gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
+-    gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
++    # gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
++    # gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
+     cd ../../../../..
+ 
+     echo -e "${GREEN}Installation completed successfully!${NC}"
+diff --git a/train_all.sh b/train_all.sh
+index 0157306..33035f5 100755
+--- a/train_all.sh
++++ b/train_all.sh
+@@ -6,246 +6,257 @@ USE_GRPO="algorithm.adv_estimator=grpo"
+ USE_PPO="algorithm.adv_estimator=gae" # by default.
+ USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+ 
++
++# python train.py --config-name _8_2048 system.CUDA_VISIBLE_DEVICES="'0,1,2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=game_2048 $USE_PPO $USE_BASE 
++
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-1 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-1.5B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'4,5'" trainer.project_name=ragen_latest_qwen_25_3b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-3B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-3 $USE_PPO $USE_BASE 
++
+ # Section 3.1&3.2 - General Observations
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozen_lake-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozen_lake-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=bandit-ppo-multitask $USE_PPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'7'" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-ppo-frommlp $USE_PPO $USE_BASE 
+ 
+-# Section 4.1 - Filtering and critic
+-# 0.25
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-
+-wait
+-
+-# 0.5
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-
+-# 0.75
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-
+-wait
+-
+-# Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=sokoban-ppo-box1-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-wait
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.project_name=ragen_latest_qwen_25_3b_it  trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_3B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-# Section 5.1 - Reasoning Helps Generalization
++# # Section 4.1 - Filtering and critic
++# # 0.25
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
+-    agent_proxy.enable_think=False
++# # 0.5
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++
++# # 0.75
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
+-    agent_proxy.enable_think=False
++# # Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+ 
++# wait
+ 
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
++# # Section 5.1 - Reasoning Helps Generalization
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
+ 
+-# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
++#     agent_proxy.enable_think=False
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader"
+ 
+-# COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
++#     agent_proxy.enable_think=False
+ 
+-wait
+ 
+-# Section 5.2 - what leads to better reasoning?
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
++# # SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
++# # COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
+ 
++# wait
+ 
+-wait
++# # Section 5.2 - what leads to better reasoning?
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
+ 
+-# Section 6.1 varying action count
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
++# wait
+ 
+-# section 6.2 Varying prompt diversity
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
+ 
+-wait 
++# wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
++# wait
+ 
++# # Section 6.1 varying action count
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
+ 
++# # section 6.2 Varying prompt diversity
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
+ 
++# wait 
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
+ 
+-# Extension: Training 7B reasoning model
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
+ 
+-# trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
++# # Extension: Training 7B reasoning model
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
+ 
++# python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
+ 
++# # trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_PPO="algorithm.adv_estimator=gae" # by default.
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
+-# enable_response_mask: False
+-# grpo_advantage_length_weight: True
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+ 
+-# extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
++# # USE_PPO="algorithm.adv_estimator=gae" # by default.
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
+ 
+-# extension: bi-level gae
+-python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
+-    system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
+-    model_path=Qwen/Qwen2.5-0.5B-Instruct \
+-    algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
+-    agent_proxy.use_turn_scores=True \
+-    actor_rollout_ref.rollout.tp_size_check=False
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
+ 
+-# extension: webshop
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # enable_response_mask: False
++# # grpo_advantage_length_weight: True
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: bi-level gae
++# python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
++#     system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
++#     model_path=Qwen/Qwen2.5-0.5B-Instruct \
++#     algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
++#     agent_proxy.use_turn_scores=True \
++#     actor_rollout_ref.rollout.tp_size_check=False
+ 
+ 
+-# normal:sokoban
+-# extension: sokoban
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # extension: webshop
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
+ 
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++
++# # normal:sokoban
++# # extension: sokoban
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
++
++
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
++
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+ 
+ 
+-python train.py \
+-    trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+
++# python train.py \
++#     trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+

wandb/run-20251218_182448-on6j4c96/files/diff_8d73639b99b38265453f898b8d6af7d4af50d56e.patch

@@ -0,0 +1,1162 @@
+diff --git a/config/base.yaml b/config/base.yaml
+index a1659e4..9df6250 100644
+--- a/config/base.yaml
++++ b/config/base.yaml
+@@ -9,15 +9,16 @@ seed:
+   train: 10000
+   val: 123
+ 
+-micro_batch_size_per_gpu: 4
++micro_batch_size_per_gpu: 1
+ ppo_mini_batch_size: 32
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/Qwen/Qwen2.5-0.5B-Instruct
++# Qwen/Qwen2.5-0.5B-Instruct
+ enable_response_mask: True # Enabling response mask could improve stability of rollout/old_log_prob, as P(st|history) are no longer calculated in loss here. See https://docs.google.com/document/d/1bg7obeiKTExuHHBl5uOiSpec5uLDZ2Tgvxy6li5pHX4/edit?usp=sharing for more details.
+ grpo_advantage_length_weight: False # if you do not enable this and critic/advantage_estimator is GRPO, and the critic/advantages/mean is too low, then you can try enabling this to encourage reasoning and forbid collapse
+ 
+ lora:
+   rank: 0
+-  alpha: 64
++  alpha: 16
+   target_modules: all-linear
+ 
+ actor_rollout_ref:
+@@ -46,10 +47,10 @@ actor_rollout_ref:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu} # following micro_batch_size_per_gpu
+     tensor_model_parallel_size: 1
+-    max_model_len: 3600
++    max_model_len: 7200 #3600 why** 14400
+     prompt_length: 1 # useless. Just put it here
+     response_length: 400 # single-turn response length
+-    gpu_memory_utilization: 0.5
++    gpu_memory_utilization: 0.7
+     max_num_batched_tokens: 8192 # set only when enable_chunked_prefill is true
+     temperature: 1
+     rollout_filter_ratio: 0.25
+@@ -90,26 +91,28 @@ algorithm:
+     kl_coef: 0.000
+ 
+ trainer:
+-  project_name: ragen_latest
++  project_name: 
+   experiment_name: test
+   local_log_dir: "results/"
+-  save_freq: 100
++  save_freq: -1
+   total_training_steps: 200
+   validation_steps: 1 # validation instances = validation_steps * val_env_groups * group_size
+   val_before_train: True
+   n_gpus_per_node: 1
+   test_freq: 10
+   generations_to_log_to_wandb: 
++    train: 128
+     val: 20
+   logger: [ 'console', 'wandb' ]
+   max_actor_ckpt_to_keep: 1
+   max_critic_ckpt_to_keep: 1
++  default_local_dir: /mnt/general/wanghy/RAGEN/saves/
+ 
+ agent_proxy:
+   max_context_window: -1 # set a value > 0 to enable context window for long trajectory
+-  max_turn: 5
++  max_turn: 25 #25 why** 700
+   action_sep: "||"
+-  max_actions_per_turn: 2 # how many actions can be output at most in a single turn
++  max_actions_per_turn: 1 # how many actions can be output at most in a single turn
+   use_turn_scores: False # important to GAE when applying token-level rewards to token-level advantages. If False, will take the sum of scores as the reward for the last turn.
+   enable_think: True # False -> no think RL
+   reward_normalization:
+@@ -126,11 +129,11 @@ es_manager:
+       tags: ["CoordSokoban"]
+       n_groups: [8] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+   val:
+-    env_groups: 32
+-    group_size: 16 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
++    env_groups: 256
++    group_size: 1 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
+     env_configs:
+       tags: ["CoordSokoban"]
+-      n_groups: [32] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      n_groups: [256] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ ctx_manager:
+   generation: # go to vllm
+diff --git a/config/envs.yaml b/config/envs.yaml
+index 1002ac8..ba3ac6f 100644
+--- a/config/envs.yaml
++++ b/config/envs.yaml
+@@ -2,6 +2,7 @@ custom_envs:
+   SimpleSokoban:
+     env_type: sokoban
+     max_actions_per_traj: 10 # used in environment state manager to control the actual max actions executed per trajectory
++    # ORIGNAL env_instruction
+     env_instruction: |
+       You are solving the Sokoban puzzle. 
+       You are the player and you need to push all boxes to targets. 
+@@ -12,7 +13,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+ 
+   LargerSokoban:
+@@ -65,7 +66,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+       observation_format: "grid_coord"
+ 
+@@ -111,14 +112,14 @@ custom_envs:
+ 
+   FrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: "You are solving the FrozenLake puzzle. Forbid the whole and go to the target. You may move to the unintended direction due to the slippery ice. Example answer format: <think>To forbid the hole and go to the target, I should go left then go up.</think><answer>Left || Up</answer>"
+     max_tokens: 100
+     env_config: null
+ 
+   CoordFrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: |
+       You are solving the FrozenLake puzzle. The observation includes both a symbol grid and zero-indexed coordinates for the start, goal, player, and any holes.
+       Coordinates range from the top-left corner (0, 0) to the bottom-right corner (5, 5).
+@@ -195,4 +196,80 @@ custom_envs:
+     max_actions_per_traj: 30
+     env_instruction: "You are a Lean theorem prover. Given a Lean theorem statement, propose a sequence of tactics that completes the proof. Think step by step about which tactics to apply next. Provide tactics separated by '||', for example <answer>intro || simp || rfl</answer>."
+     max_tokens: 512
+-    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
+
++    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
++
++  
++  game_2048:
++    env_type: game_2048
++    max_actions_per_traj: 700
++    env_instruction: |
++      You are playing the 2048 game on a 4x4 grid. Merge equal tiles by sliding Up, Right, Down, or Left.
++      If a move is invalid (no tiles move), a small penalty is applied. Respond with a single action.
++      Example: <answer>Up</answer>
++    max_tokens: 8192
++    env_config: null
++
++  blackjack:
++    env_type: blackjack
++    max_actions_per_traj: 10
++    env_instruction: |
++      You are playing Blackjack against a dealer. The dealer must hit on 16 or less and stand on 17 or more.
++      Choose either Stick or Hit. Respond with a single action.
++      Example: <answer>Hit</answer>
++    max_tokens: 64
++    env_config: null
++
++  rubikscube:
++    env_type: rubikscube
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.
++      Available actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.
++      - Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).
++      - Modifiers: A letter alone means 90° clockwise (e.g., 'R'). A letter with prime (') means 90° counter-clockwise (e.g., "R'").
++      Respond with a sequence of actions separated by "||".
++      Example: <answer>U</answer>
++    max_tokens: 96
++    env_config:
++      scramble_depth: 1
++      max_steps: 20
++      render_mode: "text"
++  
++  SimpleSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 2x2 box contains the numbers 1-4 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format, for example: <answer>place 1 at row 2 col 3</answer> or <answer>1,2,3</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 4
++      difficulty: "easy"
++      render_format: "with_feedback"
++      show_conflicts: false
++      show_valid_numbers: false
++      max_steps: 20
++
++  MediumSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 30
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 3x3 box contains the numbers 1-9 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format: <answer>place 5 at row 2 col 3</answer> or <answer>2,3,5</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 9
++      difficulty: "medium"
++      render_format: "with_feedback"
++      show_conflicts: true
++      show_valid_numbers: true
++      max_steps: 81
++
+diff --git a/config/eval.yaml b/config/eval.yaml
+index e65f3a5..7775fd1 100644
+--- a/config/eval.yaml
++++ b/config/eval.yaml
+@@ -1,14 +1,15 @@
+ defaults:
+   - envs
+-
++enable_response_mask: True
+ system:
+-  CUDA_VISIBLE_DEVICES: "0"
++  CUDA_VISIBLE_DEVICES: "0,1,2,3,4,5,6,7"
+ 
+ seed:
+   train: 10000
+   val: 123
+ 
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/openai/gpt-oss-20b
++# /mnt/general/share/model/tyzhu/SPA-frozenlake-qwen2.5-1.5b-instruct
+ 
+ lora:
+   rank: 0
+@@ -24,9 +25,9 @@ actor_rollout_ref:
+   rollout:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu}
+-    tensor_model_parallel_size: 1
++    tensor_model_parallel_size: 8
+     dtype: bfloat16
+-    max_model_len: 3600
++    max_model_len: 7200
+     prompt_length: 1
+     response_length: 400
+     gpu_memory_utilization: 0.9
+@@ -35,6 +36,7 @@ actor_rollout_ref:
+     free_cache_engine: True
+     enable_chunked_prefill: False
+     disable_log_stats: False
++    do_sample: True
+     val_kwargs:
+       do_sample: True
+       temperature: 0.5
+@@ -44,9 +46,9 @@ actor_rollout_ref:
+ 
+ agent_proxy:
+   max_context_window: -1
+-  max_turn: 5
++  max_turn: 25
+   action_sep: "||"
+-  max_actions_per_turn: 2
++  max_actions_per_turn: 1
+   use_turn_scores: False
+   enable_think: True
+   reward_normalization:
+@@ -59,13 +61,13 @@ es_manager:
+     env_groups: 8
+     group_size: 16
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [8]
+   val:
+     env_groups: 32
+-    group_size: 16
++    group_size: 128
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [32]
+ 
+ ctx_manager:
+diff --git a/config/evaluate_api_llm.yaml b/config/evaluate_api_llm.yaml
+index b21c95e..e2e6761 100644
+--- a/config/evaluate_api_llm.yaml
++++ b/config/evaluate_api_llm.yaml
+@@ -1,8 +1,11 @@
++#export OPENAI_BASE_URL="https://api.ohmygpt.com/v1"
++#export OPENAI_API_KEY="sk-o4sMxBkN5BB100C4D4a3T3BlBkFJF7791CA39EA14ca98041"
++#python -m ragen.eval_api hydra.searchpath='[file://./verl/verl/trainer/config]'
+ defaults:
+   - base # this is a symbolic link to the verl/verl/trainer/config/ppo_trainer.yaml file
+ 
+ model_config:
+-  model_name: gpt-4o # should be registered in model_info
++  model_name: ark-deepseek-v3-250324 # should be registered in model_info
+   max_concurrency: 16
+ 
+ model_info:
+@@ -24,9 +27,9 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+-  gpt-4o:
++  gpt-4o-mini:
+     provider_name: openai
+-    model_name: gpt-4o
++    model_name: gpt-4o-mini
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+@@ -36,21 +39,42 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  ark-deepseek-v3-250324:
++    provider_name: openai
++    model_name: ark-deepseek-v3-250324
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
+   deepseek-v3:
+     provider_name: deepseek
+     model_name: deepseek-chat
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  glm-4.6:
++    provider_name: openai
++    model_name: glm-4.6
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
++  TA/openai/gpt-oss-120b:
++    provider_name: openai
++    model_name: TA/openai/gpt-oss-120b
++    generation_kwargs:
++      temperature: 0
++      max_tokens: 8192
++      # max_retries: 5
+ 
+-
+-
++agent_proxy:
++  max_turn: 5
+ es_manager:
+   val:
+-    env_groups: 256
++    env_groups: 128
+     group_size: 1 # should be set to 1 because val temperature is set to 0 and same prompt leads to same output
+     env_configs:
+-      tags: ["CoordSokoban"]
+-      n_groups: [256] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      tags: ["rubikscube"]
++      n_groups: [128] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ 
++rollout:
++  max_model_len: 7200
+
+Submodule external/webshop-minimal contains modified content
+diff --git a/external/webshop-minimal/requirements.txt b/external/webshop-minimal/requirements.txt
+index 5a1b04f..238ed5a 100644
+--- a/external/webshop-minimal/requirements.txt
++++ b/external/webshop-minimal/requirements.txt
+@@ -4,7 +4,7 @@ flask
+ html2text
+ rank_bm25
+ pyserini
+-faiss-cpu
++faiss-gpu
+ thefuzz
+ gdown
+ spacy
+diff --git a/ragen/env/__init__.py b/ragen/env/__init__.py
+index b0f3461..bc6838e 100644
+--- a/ragen/env/__init__.py
++++ b/ragen/env/__init__.py
+@@ -12,6 +12,14 @@ from .metamathqa.env import MetaMathQAEnv
+ from .metamathqa.config import MetaMathQAEnvConfig
+ from .lean.config import LeanEnvConfig
+ from .lean.env import LeanEnv
++from .game_2048.config import Game2048EnvConfig
++from .game_2048.env import Game2048Env
++from .blackjack.config import BlackjackEnvConfig
++from .blackjack.env import BlackjackEnv
++from .rubikscube.config import RubiksCube2x2Config
++from .rubikscube.env import RubiksCube2x2Env
++from .sudoku.config import SudokuEnvConfig
++from .sudoku.env import SudokuEnv
+ 
+ 
+ REGISTERED_ENVS = {
+@@ -22,6 +30,10 @@ REGISTERED_ENVS = {
+     # 'alfworld': AlfredTXTEnv,
+     'metamathqa': MetaMathQAEnv,
+     'lean': LeanEnv,
++    'game_2048': Game2048Env,
++    'blackjack': BlackjackEnv,
++    'rubikscube': RubiksCube2x2Env,
++    'sudoku': SudokuEnv,
+ }
+ 
+ REGISTERED_ENV_CONFIGS = {
+@@ -32,6 +44,10 @@ REGISTERED_ENV_CONFIGS = {
+     # 'alfworld': AlfredEnvConfig,
+     'metamathqa': MetaMathQAEnvConfig,
+     'lean': LeanEnvConfig,
++    'game_2048': Game2048EnvConfig,
++    'blackjack': BlackjackEnvConfig,
++    'rubikscube': RubiksCube2x2Config,
++    'sudoku': SudokuEnvConfig,
+ }
+ 
+ try:
+diff --git a/ragen/env/frozen_lake/config.py b/ragen/env/frozen_lake/config.py
+index de054f4..9950c34 100644
+--- a/ragen/env/frozen_lake/config.py
++++ b/ragen/env/frozen_lake/config.py
+@@ -8,7 +8,7 @@ class FrozenLakeEnvConfig:
+     size: int = 4
+     p: float = 0.9
+     success_rate: float = 0.8
+-    is_slippery: bool = True
++    is_slippery: bool = False
+     map_seed: Optional[int] = None
+     render_mode: str = "text"
+     observation_format: str = "grid"
+diff --git a/ragen/env/frozen_lake/env.py b/ragen/env/frozen_lake/env.py
+index 9ee3add..eef11d8 100644
+--- a/ragen/env/frozen_lake/env.py
++++ b/ragen/env/frozen_lake/env.py
+@@ -13,6 +13,7 @@ from ragen.env.base import BaseDiscreteActionEnv
+ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+     def __init__(self, config: FrozenLakeEnvConfig = FrozenLakeEnvConfig()):
+         # Using mappings directly from config
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.GRID_LOOKUP = config.grid_lookup
+         self.ACTION_LOOKUP = config.action_lookup
+@@ -95,6 +96,7 @@ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+ if __name__ == "__main__":
+     import matplotlib.pyplot as plt
+     config = FrozenLakeEnvConfig(size=4, is_slippery=True)
++    # import pdb;pdb.set_trace()
+     env = FrozenLakeEnv(config)
+     print(env.reset())
+     while True:
+diff --git a/ragen/env/sokoban/env.py b/ragen/env/sokoban/env.py
+index 17cd636..6922b54 100644
+--- a/ragen/env/sokoban/env.py
++++ b/ragen/env/sokoban/env.py
+@@ -33,6 +33,7 @@ class SokobanEnv(BaseDiscreteActionEnv, GymSokobanEnv):
+     def reset(self, seed=None, mode=None):
+         try:
+             with all_seed(seed):
++                # import pdb;pdb.set_trace()
+                 self.room_fixed, self.room_state, self.box_mapping, action_sequence = generate_room(
+                     dim=self.dim_room,
+                     num_steps=self.num_gen_steps,
+diff --git a/ragen/llm_agent/agent_proxy.py b/ragen/llm_agent/agent_proxy.py
+index c25b50d..f6c4088 100644
+--- a/ragen/llm_agent/agent_proxy.py
++++ b/ragen/llm_agent/agent_proxy.py
+@@ -287,6 +287,7 @@ def main(config):
+ 	proxy = LLMAgentProxy(config, actor_wg, tokenizer)
+ 	import time
+ 	start_time = time.time()
++	# import pdb;pdb.set_trace()
+ 	rollouts = proxy.rollout(
+ 		DataProto(
+ 			batch=None,
+diff --git a/ragen/llm_agent/base_llm.py b/ragen/llm_agent/base_llm.py
+index 358a2eb..b6f7797 100644
+--- a/ragen/llm_agent/base_llm.py
++++ b/ragen/llm_agent/base_llm.py
+@@ -33,6 +33,7 @@ class OpenAIProvider(LLMProvider):
+             raise ValueError("OpenAI API key not provided and not found in environment variables")
+         
+         self.client = AsyncOpenAI(api_key=self.api_key)
++        # import pdb;pdb.set_trace()
+     
+     async def generate(self, messages: List[Dict[str, str]], **kwargs) -> LLMResponse:
+         if "o1-mini" in self.model_name:
+@@ -46,10 +47,7 @@ class OpenAIProvider(LLMProvider):
+         )
+         if response.choices[0].finish_reason in ['length', 'content_filter']:
+             raise ValueError("Content filtered or length exceeded")
+-        return LLMResponse(
+-            content=response.choices[0].message.content,
+-            model_name=response.model
+-        )
++        return LLMResponse(content=response.choices[0].message.content,model_name=response.model)
+ 
+ class DeepSeekProvider(LLMProvider):
+     """DeepSeek API provider implementation"""
+@@ -199,7 +197,7 @@ class ConcurrentLLM:
+         
+         # Queue to store unfinished or failed tasks
+         current_batch = messages_list.copy()
+-        max_retries = kwargs.get("max_retries", 100)
++        max_retries = kwargs.get("max_retries", 10)
+         retry_count = 0
+         
+         while current_batch and retry_count < max_retries:
+diff --git a/ragen/llm_agent/ctx_manager.py b/ragen/llm_agent/ctx_manager.py
+index 905247a..af20a01 100644
+--- a/ragen/llm_agent/ctx_manager.py
++++ b/ragen/llm_agent/ctx_manager.py
+@@ -88,6 +88,7 @@ class ContextManager:
+         Initialize the ContextManager.
+         Processor is used to process the image data.
+         """
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.tokenizer = tokenizer
+         self.processor = processor
+@@ -311,7 +312,7 @@ class ContextManager:
+ 
+             llm_input_texts.append(text_with_prompt)
+             messages_list.append(messages)
+-
++        # import pdb;pdb.set_trace()
+         inputs = self.tokenizer(llm_input_texts, return_tensors="pt", padding=True, padding_side="left", truncation=False) # We have truncated previously, truncation in tokenizer may cause issues.
+         input_ids, attention_mask = inputs.input_ids, inputs.attention_mask
+         position_ids = (attention_mask.cumsum(dim=-1) - 1).clamp(min=0)
+@@ -353,6 +354,30 @@ class ContextManager:
+                 key: np.sum(value) / self.env_nums[key.split("/")[0]]
+                 for key, value in metrics.items()
+             }
++            # Derived metrics for wandb logging
++            try:
++                # charts/avg_episode_return: average across all 2048 env groups
++                ep_keys = [k for k in metrics.keys() if k.endswith('/episodic_return')]
++                if len(ep_keys) > 0:
++                    ep_vals = []
++                    for k in ep_keys:
++                        tag = k.split('/')[0]
++                        denom = self.env_nums.get(tag, max(1, len(metrics[k])))
++                        ep_vals.append(float(np.sum(metrics[k]) / denom))
++                    mean_metrics["charts/avg_episode_return"] = float(np.mean(ep_vals))
++            except Exception:
++                pass
++            try:
++                # rollout/max_tile: max over all envs in this batch
++                tile_keys = [k for k in metrics.keys() if k.endswith('/max_tile')]
++                if len(tile_keys) > 0:
++                    tile_vals = []
++                    for k in tile_keys:
++                        tile_vals.extend(metrics[k])
++                    if len(tile_vals) > 0:
++                        mean_metrics["rollout/max_tile"] = int(np.max(tile_vals))
++            except Exception:
++                pass
+             for key, values in metrics.items():
+                 if not isinstance(values, list):
+                     continue
+diff --git a/ragen/llm_agent/es_manager.py b/ragen/llm_agent/es_manager.py
+index b87a2b3..1a49d3e 100644
+--- a/ragen/llm_agent/es_manager.py
++++ b/ragen/llm_agent/es_manager.py
+@@ -128,18 +128,40 @@ class EnvStateManager:
+         env_outputs: List[Dict]
+             {env_id: int, history: List[Dict][{state: str, actions: List[str], reward: float, info: Dict, llm_response: str, llm_raw_response: str, (Optional)images: List[PIL.Image.Image]}]}
+         """
++        # def _execute_actions(env, actions):
++        #     acc_reward, turn_info, turn_done = 0, {}, False
++        #     executed_actions = []
++        #     for action in actions:
++        #         _, reward, done, info = env.step(action)
++        #         acc_reward += reward
++        #         turn_info.update(info) # NOTE: currently use last info for multi-action
++        #         executed_actions.append(action)
++        #         if done:
++        #             turn_done = True
++        #             break
++            
++        #     return acc_reward, turn_info, turn_done, executed_actions
+         def _execute_actions(env, actions):
+-            acc_reward, turn_info, turn_done = 0, {}, False
++            acc_reward, turn_info, turn_done = 0.0, {}, False
++            raw_acc_reward = 0.0
+             executed_actions = []
+             for action in actions:
+                 _, reward, done, info = env.step(action)
+-                acc_reward += reward
++                acc_reward += float(reward)
++                try:
++                    raw_acc_reward += float(info.get('raw_reward', 0.0))
++                except Exception:
++                    pass
+                 turn_info.update(info) # NOTE: currently use last info for multi-action
+                 executed_actions.append(action)
+                 if done:
+                     turn_done = True
+                     break
+-            
++            # Overwrite per-turn raw_reward to reflect the sum across all executed actions in this turn
++            try:
++                turn_info['raw_reward'] = float(raw_acc_reward)
++            except Exception:
++                pass
+             return acc_reward, turn_info, turn_done, executed_actions
+ 
+         def _log_env_state(status, history, cur_obs, max_actions_per_traj, executed_actions, all_actions, acc_reward, turn_done, turn_info, env_input):
+@@ -198,6 +220,18 @@ class EnvStateManager:
+                 'success': float(status.terminated and (not status.truncated)),
+                 'num_actions': status.num_actions,
+             }
++            # Add episodic-level metrics
++            # try:
++            #     # Sum of per-turn rewards equals the episodic return (env-shaped reward)
++            #     env_metric['episodic_return'] = float(sum(status.rewards))
++            # except Exception:
++            #     pass
++            try:
++                # Final max tile on the board at the end of the rollout
++                import numpy as _np
++                env_metric['max_tile'] = int(_np.max(entry['env'].grid))
++            except Exception:
++                pass
+             custom_metric = {}
+             for turn in cache['history']:
+                 for k, v in turn.get('info', {}).items():
+@@ -212,6 +246,12 @@ class EnvStateManager:
+                             "Skipping non-numeric metric '%s' with value %r for env %s.",
+                             k, v, entry['tag']
+                         )
++            # Add episodic_return as the SUM of raw_reward across steps (align with CleanRL)
++            try:
++                if 'raw_reward' in custom_metric:
++                    env_metric['episodic_return'] = float(np.sum(custom_metric['raw_reward']))
++            except Exception:
++                pass
+             for k, v in custom_metric.items():
+                 # TODO: Move TURN_LVL_METRICS into the environment
+                 if "webshop" not in cache['tag'].lower() or ("webshop" in cache['tag'].lower() and k in TURN_LVL_METRICS):
+@@ -219,7 +259,12 @@ class EnvStateManager:
+                 else:
+                     env_metric['traj_sum/' + k] = np.sum(v)
+ 
+-
++            try:
++                if 'score' in custom_metric and len(custom_metric['score']) > 0:
++                    env_metric['final_score'] = float(custom_metric['score'][-1])
++            except Exception:
++                pass
++                
+             cache['history'][-1]['metrics'] = custom_metric
+             env_metric = {f"{entry['tag']}/{k}": v for k, v in env_metric.items()}
+             cache['metrics'] = env_metric
+diff --git a/requirements.txt b/requirements.txt
+index 2fd756e..9b521bf 100644
+--- a/requirements.txt
++++ b/requirements.txt
+@@ -7,7 +7,6 @@ accelerate
+ codetiming
+ datasets
+ dill
+-flash-attn==2.7.4.post1
+ hydra-core
+ numpy
+ pandas
+@@ -15,19 +14,19 @@ pybind11
+ ray>=2.10
+ tensordict>=0.8.0,<0.9.0
+ transformers
+-vllm==0.8.2
++vllm==0.8.5
+ wandb
+ gymnasium
+ gymnasium[toy-text]
+ 
+ pyarrow>=15.0.0
+ pylatexenc
+-torchdata
++# torchdata
+ debugpy
+ 
+ together
+ anthropic
+-faiss-cpu==1.11.0
++faiss-gpu
+ 
+ # This is optional, but needs to be installed with main requirements if you want to use webshop
+ -r external/webshop-minimal/requirements.txt
+diff --git a/scripts/setup_ragen.sh b/scripts/setup_ragen.sh
+index f9a7cd9..85a93c4 100644
+--- a/scripts/setup_ragen.sh
++++ b/scripts/setup_ragen.sh
+@@ -94,10 +94,10 @@ main() {
+         pip install torch==2.5.0 --index-url https://download.pytorch.org/whl/cu124
+         
+         print_step "Installing flash-attention..."
+-        pip3 install flash-attn==2.7.4.post1 --no-build-isolation
++        # pip3 install flash-attn==2.7.4.post1 --no-build-isolation
+     else
+         print_step "Installing PyTorch without CUDA support..."
+-        pip install torch==2.5.0
++        pip install torch==2.4.0
+     fi
+     
+     # Install remaining requirements
+@@ -137,8 +137,8 @@ main() {
+     conda install conda-forge::gdown
+     mkdir -p external/webshop-minimal/webshop_minimal/data/full
+     cd external/webshop-minimal/webshop_minimal/data/full
+-    gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
+-    gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
++    # gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
++    # gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
+     cd ../../../../..
+ 
+     echo -e "${GREEN}Installation completed successfully!${NC}"
+diff --git a/train_all.sh b/train_all.sh
+index 0157306..33035f5 100755
+--- a/train_all.sh
++++ b/train_all.sh
+@@ -6,246 +6,257 @@ USE_GRPO="algorithm.adv_estimator=grpo"
+ USE_PPO="algorithm.adv_estimator=gae" # by default.
+ USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+ 
++
++# python train.py --config-name _8_2048 system.CUDA_VISIBLE_DEVICES="'0,1,2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=game_2048 $USE_PPO $USE_BASE 
++
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-1 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-1.5B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'4,5'" trainer.project_name=ragen_latest_qwen_25_3b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-3B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-3 $USE_PPO $USE_BASE 
++
+ # Section 3.1&3.2 - General Observations
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozen_lake-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozen_lake-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=bandit-ppo-multitask $USE_PPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'7'" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-ppo-frommlp $USE_PPO $USE_BASE 
+ 
+-# Section 4.1 - Filtering and critic
+-# 0.25
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-
+-wait
+-
+-# 0.5
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-
+-# 0.75
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-
+-wait
+-
+-# Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=sokoban-ppo-box1-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-wait
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.project_name=ragen_latest_qwen_25_3b_it  trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_3B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-# Section 5.1 - Reasoning Helps Generalization
++# # Section 4.1 - Filtering and critic
++# # 0.25
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
+-    agent_proxy.enable_think=False
++# # 0.5
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++
++# # 0.75
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
+-    agent_proxy.enable_think=False
++# # Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+ 
++# wait
+ 
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
++# # Section 5.1 - Reasoning Helps Generalization
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
+ 
+-# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
++#     agent_proxy.enable_think=False
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader"
+ 
+-# COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
++#     agent_proxy.enable_think=False
+ 
+-wait
+ 
+-# Section 5.2 - what leads to better reasoning?
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
++# # SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
++# # COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
+ 
++# wait
+ 
+-wait
++# # Section 5.2 - what leads to better reasoning?
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
+ 
+-# Section 6.1 varying action count
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
++# wait
+ 
+-# section 6.2 Varying prompt diversity
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
+ 
+-wait 
++# wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
++# wait
+ 
++# # Section 6.1 varying action count
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
+ 
++# # section 6.2 Varying prompt diversity
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
+ 
++# wait 
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
+ 
+-# Extension: Training 7B reasoning model
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
+ 
+-# trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
++# # Extension: Training 7B reasoning model
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
+ 
++# python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
+ 
++# # trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_PPO="algorithm.adv_estimator=gae" # by default.
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
+-# enable_response_mask: False
+-# grpo_advantage_length_weight: True
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+ 
+-# extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
++# # USE_PPO="algorithm.adv_estimator=gae" # by default.
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
+ 
+-# extension: bi-level gae
+-python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
+-    system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
+-    model_path=Qwen/Qwen2.5-0.5B-Instruct \
+-    algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
+-    agent_proxy.use_turn_scores=True \
+-    actor_rollout_ref.rollout.tp_size_check=False
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
+ 
+-# extension: webshop
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # enable_response_mask: False
++# # grpo_advantage_length_weight: True
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: bi-level gae
++# python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
++#     system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
++#     model_path=Qwen/Qwen2.5-0.5B-Instruct \
++#     algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
++#     agent_proxy.use_turn_scores=True \
++#     actor_rollout_ref.rollout.tp_size_check=False
+ 
+ 
+-# normal:sokoban
+-# extension: sokoban
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # extension: webshop
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
+ 
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++
++# # normal:sokoban
++# # extension: sokoban
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
++
++
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
++
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+ 
+ 
+-python train.py \
+-    trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+
++# python train.py \
++#     trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+

wandb/run-20251218_182448-on6j4c96/files/requirements.txt

@@ -0,0 +1,315 @@
+colorama==0.4.6
+psutil==7.1.2
+ragen==0.1
+ragen==0.1
+setuptools==80.9.0
+wheel==0.45.1
+pip==25.3
+ragen==0.1
+pytz==2025.2
+nvidia-cusparselt-cu12==0.6.2
+mpmath==1.3.0
+typing_extensions==4.15.0
+sympy==1.13.1
+pylatexenc==2.10
+nvidia-cusolver-cu12==11.6.1.9
+numpy==1.26.4
+Markdown==3.10
+absl-py==2.3.1
+pyasn1_modules==0.4.2
+Werkzeug==3.1.3
+networkx==3.3
+MarkupSafe==2.1.5
+fsspec==2025.9.0
+filelock==3.19.1
+nvidia-cudnn-cu12==9.1.0.70
+Jinja2==3.1.6
+tensorboard==2.20.0
+triton==3.2.0
+pillow==11.3.0
+xxhash==3.6.0
+nvidia-nvtx-cu12==12.4.127
+nvidia-nvjitlink-cu12==12.4.127
+nvidia-nccl-cu12==2.21.5
+nvidia-curand-cu12==10.3.5.147
+wrapt==2.0.1
+nvidia-cufft-cu12==11.2.1.3
+pyjnius==1.7.0
+nvidia-cuda-runtime-cu12==12.4.127
+nvidia-cuda-nvrtc-cu12==12.4.127
+websockets==15.0.1
+nvidia-cuda-cupti-cu12==12.4.127
+zipp==3.23.0
+nvidia-cublas-cu12==12.4.5.8
+nvidia-cusparse-cu12==12.3.1.170
+flash_attn==2.8.3
+rsa==4.9.1
+einops==0.8.1
+py-cpuinfo==9.0.0
+pure_eval==0.2.3
+ptyprocess==0.7.0
+gym-notices==0.1.0
+flatbuffers==25.9.23
+fastrlock==0.8.3
+Farama-Notifications==0.0.4
+faiss-gpu==1.7.2
+antlr4-python3-runtime==4.9.3
+wcwidth==0.2.14
+wasabi==1.1.3
+uvloop==0.22.1
+urllib3==2.5.0
+tzdata==2025.2
+typing-inspection==0.4.2
+traitlets==5.14.3
+tqdm==4.67.1
+tomli==2.3.0
+threadpoolctl==3.6.0
+tenacity==9.1.2
+tabulate==0.9.0
+spacy-loggers==1.0.5
+spacy-legacy==3.0.12
+soupsieve==2.8
+sniffio==1.3.1
+smmap==5.0.2
+six==1.17.0
+shellingham==1.5.4
+sentencepiece==0.2.1
+scipy==1.15.3
+safetensors==0.6.2
+rpds-py==0.29.0
+rignore==0.7.6
+regex==2025.11.3
+RapidFuzz==3.14.3
+rank-bm25==0.2.2
+pyzmq==27.1.0
+PyYAML==6.0.3
+pytokens==0.3.0
+python-multipart==0.0.20
+python-json-logger==4.0.0
+python-dotenv==1.2.1
+PySocks==1.7.1
+pyparsing==3.2.5
+PyJWT==2.10.1
+Pygments==2.19.2
+pygame==2.6.1
+pydantic_core==2.41.5
+pycparser==2.23
+pycountry==24.6.1
+pybind11==3.0.1
+pyarrow==22.0.0
+psutil==7.1.3
+protobuf==4.25.8
+propcache==0.4.1
+prometheus_client==0.23.1
+platformdirs==4.5.0
+pexpect==4.9.0
+pathspec==0.12.1
+partial-json-parser==0.2.1.1.post6
+parso==0.8.5
+packaging==25.0
+orjson==3.11.4
+opentelemetry-semantic-conventions-ai==0.4.13
+opencv-python-headless==4.12.0.88
+ninja==1.13.0
+nest-asyncio==1.6.0
+mypy_extensions==1.1.0
+murmurhash==1.0.15
+multidict==6.7.0
+msgspec==0.19.0
+msgpack==1.1.2
+mdurl==0.1.2
+loguru==0.7.3
+llvmlite==0.44.0
+llguidance==0.7.30
+lark==1.2.2
+kiwisolver==1.4.9
+joblib==1.5.2
+jiter==0.12.0
+itsdangerous==2.2.0
+interegular==0.3.3
+ImageIO==2.37.2
+idna==3.11
+humanfriendly==10.0
+httpx-sse==0.4.3
+httptools==0.7.1
+html2text==2025.4.15
+hf-xet==1.2.0
+h11==0.16.0
+grpcio==1.76.0
+frozenlist==1.8.0
+fonttools==4.60.1
+executing==2.2.1
+exceptiongroup==1.3.0
+eval_type_backport==0.2.2
+docstring_parser==0.17.0
+dnspython==2.8.0
+distro==1.9.0
+diskcache==5.6.3
+dill==0.4.0
+decorator==5.2.1
+debugpy==1.8.17
+Cython==3.2.1
+cymem==2.0.13
+cycler==0.12.1
+cupy-cuda12x==13.6.0
+contourpy==1.3.2
+colorama==0.4.6
+codetiming==1.4.0
+cloudpickle==3.1.2
+cloudpathlib==0.23.0
+click==8.3.1
+charset-normalizer==3.4.4
+certifi==2025.11.12
+catalogue==2.0.10
+cachetools==6.2.2
+blis==1.3.2
+blinker==1.9.0
+blake3==1.0.8
+attrs==25.4.0
+async-timeout==5.0.1
+asttokens==3.0.1
+astor==0.8.1
+annotated-types==0.7.0
+annotated-doc==0.0.4
+airportsdata==20250909
+aiohappyeyeballs==2.6.1
+yarl==1.22.0
+uvicorn==0.38.0
+typer-slim==0.20.0
+thefuzz==0.22.1
+stack-data==0.6.3
+srsly==2.5.1
+smart_open==7.5.0
+sentry-sdk==2.44.0
+scikit-learn==1.7.2
+requests==2.32.5
+referencing==0.37.0
+python-dateutil==2.9.0.post0
+pydantic==2.12.4
+prompt_toolkit==3.0.52
+preshed==3.0.11
+opentelemetry-proto==1.26.0
+omegaconf==2.3.0
+numba==0.61.2
+nltk==3.9.2
+multiprocess==0.70.18
+matplotlib-inline==0.2.1
+markdown-it-py==4.0.0
+jedi==0.19.2
+importlib_metadata==8.0.0
+httpcore==1.0.9
+gymnasium==1.2.2
+gym==0.26.2
+googleapis-common-protos==1.72.0
+gitdb==4.0.12
+gguf==0.17.1
+Flask==3.1.2
+faiss-cpu==1.13.0
+email-validator==2.3.0
+depyf==0.18.0
+Deprecated==1.3.1
+coloredlogs==15.0.1
+cffi==2.0.0
+black==25.11.0
+beautifulsoup4==4.14.2
+anyio==4.11.0
+aiosignal==1.4.0
+watchfiles==1.1.1
+tiktoken==0.12.0
+starlette==0.49.3
+sse-starlette==3.0.3
+rich==14.2.0
+pydantic-settings==2.12.0
+pydantic-extra-types==2.10.6
+pandas==2.3.3
+opentelemetry-exporter-otlp-proto-common==1.26.0
+onnxruntime==1.23.2
+matplotlib==3.10.7
+lm-format-enforcer==0.10.12
+jsonschema-specifications==2025.9.1
+ipython==8.37.0
+hydra-core==1.3.2
+huggingface-hub==0.36.0
+httpx==0.28.1
+gym-sokoban==0.0.6
+GitPython==3.1.45
+cryptography==46.0.3
+confection==0.1.5
+cleantext==1.1.4
+aiohttp==3.13.2
+weasel==0.4.3
+wandb==0.23.0
+typer==0.19.2
+virtualenv==20.35.4
+pyasn1==0.6.1
+torch==2.6.0
+tokenizers==0.22.1
+thinc==8.3.9
+rich-toolkit==0.15.1
+prometheus-fastapi-instrumentator==7.1.0
+openai==2.8.0
+jsonschema==4.25.1
+gdown==5.2.0
+fastapi==0.121.2
+anthropic==0.73.0
+xformers==0.0.29.post2
+transformers==4.57.1
+proto-plus==1.26.1
+torchvision==0.21.0
+torchdata==0.11.0
+torchaudio==2.6.0
+together==1.5.30
+tensordict==0.8.3
+spacy==3.8.9
+ray==2.51.1
+outlines_core==0.1.26
+mistral_common==1.8.5
+mcp==1.21.1
+fastapi-cloud-cli==0.3.1
+fastapi-cli==0.0.16
+datasets==4.4.1
+accelerate==1.11.0
+xgrammar==0.1.18
+pyserini==1.2.0
+peft==0.18.0
+outlines==0.1.11
+opentelemetry-exporter-otlp-proto-http==1.26.0
+opentelemetry-exporter-otlp-proto-grpc==1.26.0
+kimina-client==0.2.1
+compressed-tensors==0.9.3
+opentelemetry-exporter-otlp==1.26.0
+vllm==0.8.5
+py-spy==0.4.1
+opencensus-context==0.1.3
+distlib==0.4.0
+colorful==0.5.8
+tensorboard-data-server==0.7.2
+google-auth==2.43.0
+google-api-core==2.28.1
+aiohttp-cors==0.8.1
+opentelemetry-exporter-prometheus==0.59b0
+opencensus==0.11.4
+verl==0.5.0.dev0
+opentelemetry-api==1.38.0
+opentelemetry-semantic-conventions==0.59b0
+opentelemetry-sdk==1.38.0
+typeguard==4.4.4
+shtab==1.8.0
+tyro==0.9.35
+verl==0.5.0.dev0
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2

wandb/run-20251218_182448-on6j4c96/files/wandb-metadata.json

@@ -0,0 +1,92 @@
+{
+  "os": "Linux-5.14.0-284.25.1.el9_2.x86_64-x86_64-with-glibc2.35",
+  "python": "CPython 3.10.19",
+  "startedAt": "2025-12-18T10:24:48.379419Z",
+  "args": [
+    "--track"
+  ],
+  "program": "/mnt/general/wanghy/RAGEN/cleanrl/cleanrl/ppo_rubikscube.py",
+  "codePath": "cleanrl/cleanrl/ppo_rubikscube.py",
+  "codePathLocal": "cleanrl/cleanrl/ppo_rubikscube.py",
+  "git": {
+    "remote": "https://github.com/mll-lab-nu/RAGEN.git",
+    "commit": "8d73639b99b38265453f898b8d6af7d4af50d56e"
+  },
+  "email": "haoyu-wa22@mails.tsinghua.edu.cn",
+  "root": "/mnt/general/wanghy/RAGEN",
+  "host": "pt-4d654cf4576f4d23ad3d3919f12932fe-worker-0",
+  "executable": "/root/local/miniconda3/envs/ragen/bin/python",
+  "cpu_count": 64,
+  "cpu_count_logical": 128,
+  "gpu": "NVIDIA H100 80GB HBM3",
+  "gpu_count": 8,
+  "disk": {
+    "/": {
+      "total": "5153960755200",
+      "used": "447519776768"
+    }
+  },
+  "memory": {
+    "total": "2163642122240"
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-35e2d43d-4067-82ce-90d4-def9e389bf28"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-af4135e3-88f2-e9ac-518d-502c75a85429"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-d7fdeeba-fe9b-ec03-d9f7-6724fe4266b5"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-ccc4f668-3882-5a8e-2c07-c5cd08f6f666"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-7b73c0cf-d3d5-e10c-7176-a43be1e41001"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-81b58d94-5d1f-8ec2-f9d2-fd56172ed177"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-03e8bc66-3b44-6794-49fd-5392fbdda6d1"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-b7cf0ec6-7c29-1179-dceb-09565da51890"
+    }
+  ],
+  "cudaVersion": "12.4",
+  "writerId": "ubi6kfhhs7rtnrt9cd5f40pg33djhcqy"
+}

wandb/run-20251218_182448-on6j4c96/files/wandb-summary.json

@@ -0,0 +1 @@
+{"train/learning_rate":0.00014497950819672133,"charts/avg_value":0.7843058109283447,"_timestamp":1.7660536514345365e+09,"perf/SPS":2659,"charts/SPS":2659,"rollout/success_rate":1,"train/clipfrac":0.010986328125,"losses/explained_variance":0.881138488650322,"losses/value_loss":0.0039942702278494835,"train/entropy":0.663025438785553,"losses/policy_loss":-0.015878867357969284,"train/success_rate":0.959349593495935,"train/old_approx_kl":0.0028039049357175827,"_step":421088,"rollout/ep_len_mean":2,"train/approx_kl":0.004338355734944344,"global_step":421088,"charts/progress":42.11065573770492,"train/policy_loss":-0.015878867357969284,"charts/avg_reward":0.23046875,"train/value_loss":0.0039942702278494835,"charts/train_success_rate":0.959349593495935,"_runtime":161.629565481,"_wandb":{"runtime":161},"rollout/ep_rew_mean":1}

wandb/run-20251218_182448-on6j4c96/run-on6j4c96.wandb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19130985495cd09911e289e7a029663498c64b2387f18c738806c5dd26572f88
+size 18939904

wandb/run-20251218_182745-ahjkdup0/files/code/cleanrl/cleanrl/ppo_rubikscube.py

@@ -0,0 +1,517 @@
+# PPO with small MLP for RAGEN Rubik's Cube 2x2 using the existing env (no env edits)
+import os
+import random
+import time
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Tuple, Dict, Any, List
+import json
+import re
+
+import gymnasium as gym
+import numpy as np
+import torch
+import torch.nn as nn
+import torch.optim as optim
+import tyro
+from torch.distributions.categorical import Categorical
+
+import sys
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), '../../'))
+
+from ragen.env.rubikscube.env import RubiksCube2x2Env
+from ragen.env.rubikscube.config import RubiksCube2x2Config
+
+
+class RubiksCubeWrapper(gym.Env):
+    """
+    Adapter to use ragen RubiksCube2x2Env with Gymnasium vector API.
+    - Converts text observation to one-hot vector of 24 stickers x 6 colors.
+    - Maps agent actions [0..11] to env actions [1..12].
+    - Exposes proper observation_space and action_space.
+    """
+    metadata = {"render_modes": ["rgb_array", "human", "ansi"]}
+
+    def __init__(self, env: RubiksCube2x2Env):
+        super().__init__()
+        self._env = env
+        # 24 stickers, 6 colors -> one-hot size 144
+        self._colors = ['W', 'O', 'G', 'R', 'B', 'Y']
+        self._color_to_idx = {c: i for i, c in enumerate(self._colors)}
+        self.observation_space = gym.spaces.Box(low=0.0, high=1.0, shape=(24 * len(self._colors),), dtype=np.float32)
+        self.action_space = gym.spaces.Discrete(12)
+        # precompile regex to extract 4 letters within [X, X]\n [X, X]
+        # Lines look like: "Up (U):    [W, W]\n       [W, W]"
+        self._face_pat = re.compile(r"\[(?:\s*([A-Z])\s*,\s*([A-Z])\s*\])\n\s*\[(?:\s*([A-Z])\s*,\s*([A-Z])\s*\])")
+
+    def _encode_obs(self, text_obs: str) -> np.ndarray:
+        # Extract the six face blocks in the order U, L, F, R, B, D as rendered
+        faces_order = ["Up (U):", "Left (L):", "Front (F):", "Right (R):", "Back (B):", "Down (D):"]
+        onehots: List[int] = []
+        # Build a map from header to its block text
+        lines = text_obs.splitlines()
+        # Collect blocks starting after header line and including the next line for second row
+        i = 0
+        blocks: List[str] = []
+        while i < len(lines):
+            line = lines[i]
+            for header in faces_order:
+                if line.startswith(header):
+                    # Join current line's bracketed pair and the next line (which contains the second pair)
+                    # Remove the "Header:" prefix to keep only the bracket portions
+                    content = line[len(header):].strip()
+                    next_line = lines[i + 1] if i + 1 < len(lines) else ""
+                    block = f"{content}\n{next_line}"
+                    blocks.append(block)
+                    break
+            i += 1
+        # Fallback: if regex fails, return zeros
+        if len(blocks) != 6:
+            return np.zeros(24 * len(self._colors), dtype=np.float32)
+        stickers: List[int] = []
+        for blk in blocks:
+            m = self._face_pat.search(blk)
+            if not m:
+                return np.zeros(24 * len(self._colors), dtype=np.float32)
+            # order: 0,1,2,3 per render() doc
+            c0, c1, c2, c3 = m.group(1), m.group(2), m.group(3), m.group(4)
+            stickers.extend([c0, c1, c2, c3])
+        # stickers now length 24; convert to one-hot
+        grid = np.zeros((24, len(self._colors)), dtype=np.float32)
+        for idx, ch in enumerate(stickers):
+            cidx = self._color_to_idx.get(ch, None)
+            if cidx is not None:
+                grid[idx, cidx] = 1.0
+        return grid.reshape(-1)
+
+    def reset(self, *, seed: int | None = None, options: Dict[str, Any] | None = None):
+        text_obs = self._env.reset(seed=seed)
+        obs = self._encode_obs(text_obs)
+        return obs, {}
+
+    def step(self, action: int):
+        mapped = int(action) + 1  # 0..11 -> 1..12
+        text_obs, reward, done, info = self._env.step(mapped)
+        obs = self._encode_obs(text_obs)
+        terminated = bool(done)
+        truncated = False
+        return obs, float(reward), terminated, truncated, info or {}
+
+    def render(self):
+        return self._env.render()
+
+    def close(self):
+        self._env.close()
+
+
+@dataclass
+class Args:
+    exp_name: str = os.path.basename(__file__)[: -len(".py")]
+    seed: int = 1
+    torch_deterministic: bool = True
+    cuda: bool = True
+    track: bool = True
+    wandb_project_name: str = "cleanRL"
+    wandb_entity: str | None = None
+    capture_video: bool = False
+
+    # Algorithm
+    env_id: str = "RubiksCube2x2"
+    total_timesteps: int = 1000_000
+    learning_rate: float = 2.5e-4
+    num_envs: int = 8
+    num_steps: int = 128
+    anneal_lr: bool = True
+    gamma: float = 0.99
+    gae_lambda: float = 0.95
+    num_minibatches: int = 4
+    update_epochs: int = 4
+    norm_adv: bool = True
+    clip_coef: float = 0.2
+    clip_vloss: bool = True
+    ent_coef: float = 0.01
+    vf_coef: float = 0.5
+    max_grad_norm: float = 0.5
+    target_kl: float | None = None
+
+    # Rubik specific
+    scramble_depth: int = 3
+    max_steps_env: int = 20
+
+    # runtime filled
+    batch_size: int = 0
+    minibatch_size: int = 0
+    num_iterations: int = 0
+    eval_splits: int = 2
+    eval_episodes: int = 4000
+
+
+def make_env(idx, run_name, seed, scramble_depth, max_steps_env, capture_video=False):
+    def thunk():
+        config = RubiksCube2x2Config(scramble_depth=scramble_depth, max_steps=max_steps_env, render_mode='text')
+        env = RubiksCube2x2Env(config)
+        env = RubiksCubeWrapper(env)
+        env = gym.wrappers.TimeLimit(env, max_episode_steps=max_steps_env)
+        env = gym.wrappers.RecordEpisodeStatistics(env)
+        if capture_video and idx == 0:
+            env = gym.wrappers.RecordVideo(env, f"videos/{run_name}")
+        return env
+    return thunk
+
+
+def layer_init(layer, std=np.sqrt(2), bias_const=0.0):
+    torch.nn.init.orthogonal_(layer.weight, std)
+    torch.nn.init.constant_(layer.bias, bias_const)
+    return layer
+
+
+class Agent(nn.Module):
+    def __init__(self, envs):
+        super().__init__()
+        obs_shape = int(np.array(envs.single_observation_space.shape).prod())
+        hidden = 128
+        self.critic = nn.Sequential(
+            layer_init(nn.Linear(obs_shape, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, 1), std=1.0),
+        )
+        self.actor = nn.Sequential(
+            layer_init(nn.Linear(obs_shape, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, envs.single_action_space.n), std=0.01),
+        )
+
+    def get_value(self, x):
+        return self.critic(x)
+
+    def get_action_and_value(self, x, action=None):
+        logits = self.actor(x)
+        probs = Categorical(logits=logits)
+        if action is None:
+            action = probs.sample()
+        return action, probs.log_prob(action), probs.entropy(), self.critic(x)
+
+
+if __name__ == "__main__":
+    args = tyro.cli(Args)
+    args.batch_size = int(args.num_envs * args.num_steps)
+    args.minibatch_size = int(args.batch_size // args.num_minibatches)
+    args.num_iterations = args.total_timesteps // args.batch_size
+    run_name = f"{args.env_id}__{args.exp_name}__{args.seed}__{int(time.time())}"
+
+    if args.track:
+        import wandb
+        wandb.init(
+            project=args.wandb_project_name,
+            entity=args.wandb_entity,
+            config=vars(args),
+            name=run_name,
+            monitor_gym=True,
+            save_code=True,
+        )
+        try:
+            wandb.define_metric("global_step")
+            for prefix in ["train/*", "rollout/*", "eval/*", "losses/*", "charts/*", "perf/*"]:
+                wandb.define_metric(prefix, step_metric="global_step")
+        except Exception:
+            pass
+
+    # seeding
+    random.seed(args.seed)
+    np.random.seed(args.seed)
+    torch.manual_seed(args.seed)
+    torch.backends.cudnn.deterministic = args.torch_deterministic
+
+    device = torch.device("cuda" if torch.cuda.is_available() and args.cuda else "cpu")
+
+    # envs
+    envs = gym.vector.SyncVectorEnv([
+        make_env(i, run_name, args.seed, args.scramble_depth, args.max_steps_env, args.capture_video)
+        for i in range(args.num_envs)
+    ])
+    assert isinstance(envs.single_action_space, gym.spaces.Discrete)
+
+    agent = Agent(envs).to(device)
+    optimizer = optim.Adam(agent.parameters(), lr=args.learning_rate, eps=1e-5)
+
+    # storage
+    obs = torch.zeros((args.num_steps, args.num_envs) + envs.single_observation_space.shape).to(device)
+    actions = torch.zeros((args.num_steps, args.num_envs) + envs.single_action_space.shape).to(device)
+    logprobs = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    rewards = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    dones = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    values = torch.zeros((args.num_steps, args.num_envs)).to(device)
+
+    # start
+    global_step = 0
+    start_time = time.time()
+    next_obs, _ = envs.reset(seed=args.seed)
+    next_obs = torch.Tensor(next_obs).to(device)
+    next_done = torch.zeros(args.num_envs).to(device)
+
+    # eval helper similar to FrozenLake: collect greedy eval trajectories and write metrics.json
+    def collect_eval_trajectories(agent_model, make_env_fn, n_episodes, step_tag):
+        out_dir = Path(f"runs/{run_name}/trajectories/step_{step_tag}")
+        out_dir.mkdir(parents=True, exist_ok=True)
+        out_path = out_dir / "trajectories.jsonl"
+        env = make_env_fn()
+        collected = 0
+        summary_returns = []
+        summary_success = []
+        with out_path.open("w") as f:
+            while collected < n_episodes:
+                state, _ = env.reset(seed=args.seed + 100000 + collected)
+                traj_states = [state.tolist()]
+                traj_actions = []
+                traj_rewards = []
+                traj_dones = []
+                traj_success = []
+                done = False
+                step_count = 0
+                max_eval_steps = getattr(env, '_max_episode_steps', None) or int(args.max_steps_env)
+                while not done:
+                    with torch.no_grad():
+                        logits = agent_model.actor(torch.tensor(state, dtype=torch.float32, device=device).unsqueeze(0))
+                        action = int(torch.argmax(logits, dim=1).item())
+                    next_state, reward, terminated, truncated, info = env.step(action)
+                    traj_actions.append(int(action))
+                    traj_rewards.append(float(reward))
+                    step_count += 1
+                    d = bool(terminated) or bool(truncated) or (step_count >= max_eval_steps)
+                    traj_dones.append(d)
+                    traj_success.append(bool((info or {}).get('success', False)))
+                    state = next_state
+                    traj_states.append(state.tolist())
+                    done = d
+                ep_ret = float(sum(traj_rewards))
+                ep_succ = bool(any(traj_success))
+                record = {
+                    "states": traj_states,
+                    "actions": traj_actions,
+                    "rewards": traj_rewards,
+                    "dones": traj_dones,
+                    "success": traj_success,
+                    "episode_return": ep_ret,
+                    "episode_success": ep_succ,
+                }
+                f.write(json.dumps(record) + "\n")
+                collected += 1
+                summary_returns.append(ep_ret)
+                summary_success.append(1.0 if ep_succ else 0.0)
+        env.close()
+        try:
+            metrics = {
+                "global_step": int(step_tag),
+                "episodes": int(n_episodes),
+                "success_rate": float(np.mean(summary_success)) if len(summary_success) else 0.0,
+                "avg_return": float(np.mean(summary_returns)) if len(summary_returns) else 0.0,
+                "std_return": float(np.std(summary_returns)) if len(summary_returns) else 0.0,
+            }
+            with (out_dir / "metrics.json").open("w") as mf:
+                json.dump(metrics, mf)
+        except Exception as e:
+            print(f"Warning: failed to write eval metrics: {e}")
+
+    # training loop
+    eval_every_iters = max(1, args.num_iterations // args.eval_splits)
+    for iteration in range(1, args.num_iterations + 1):
+        # Anneal LR
+        if args.anneal_lr:
+            frac = 1.0 - (iteration - 1.0) / args.num_iterations
+            lrnow = frac * args.learning_rate
+            optimizer.param_groups[0]["lr"] = lrnow
+
+        # accumulate per-iteration episode successes
+        iter_successes = []
+        for step in range(0, args.num_steps):
+            global_step += args.num_envs
+            obs[step] = next_obs
+            dones[step] = next_done
+
+            with torch.no_grad():
+                action, logprob, _, value = agent.get_action_and_value(next_obs)
+                values[step] = value.flatten()
+            actions[step] = action
+            logprobs[step] = logprob
+
+            next_obs, reward, terminations, truncations, infos = envs.step(action.cpu().numpy())
+            next_done = np.logical_or(terminations, truncations)
+            rewards[step] = torch.tensor(reward).to(device).view(-1)
+            next_obs, next_done = torch.Tensor(next_obs).to(device), torch.Tensor(next_done).to(device)
+
+            # Episode stats logging similar to FrozenLake
+            try:
+                mask = None
+                if isinstance(infos, dict):
+                    if "_episode" in infos:
+                        mask = np.asarray(infos["_episode"]).astype(bool)
+                    elif "episode" in infos and isinstance(infos["episode"], dict) and "_l" in infos["episode"]:
+                        mask = np.asarray(infos["episode"]["_l"]).astype(bool)
+                if mask is not None and np.any(mask):
+                    r_arr = np.asarray(infos.get("episode", {}).get("r", np.zeros_like(mask, dtype=float)))
+                    l_arr = np.asarray(infos.get("episode", {}).get("l", np.zeros_like(mask, dtype=int)))
+                    # prefer success from info; fallback to ep return > 0
+                    if "success" in infos:
+                        succ_arr = np.asarray(infos.get("success", np.zeros_like(mask, dtype=bool))).astype(float)
+                    else:
+                        try:
+                            succ_arr = (np.asarray(r_arr) > 0).astype(float)
+                        except Exception:
+                            succ_arr = np.zeros_like(mask, dtype=float)
+                    # collect iteration successes for training success rate
+                    try:
+                        for s in np.asarray(succ_arr)[mask]:
+                            iter_successes.append(float(s))
+                    except Exception:
+                        pass
+                    if args.track:
+                        try:
+                            import wandb
+                            log_dict = {
+                                "global_step": int(global_step),
+                                "rollout/ep_rew_mean": float(np.mean(r_arr[mask])) if np.any(mask) else None,
+                                "rollout/ep_len_mean": float(np.mean(l_arr[mask])) if np.any(mask) else None,
+                                "rollout/success_rate": float(np.mean(succ_arr[mask])) if np.any(mask) else None,
+                            }
+                            wandb.log(log_dict, step=global_step)
+                        except Exception:
+                            pass
+            except Exception:
+                pass
+
+        # GAE
+        with torch.no_grad():
+            next_value = agent.get_value(next_obs).reshape(1, -1)
+            advantages = torch.zeros_like(rewards).to(device)
+            lastgaelam = 0
+            for t in reversed(range(args.num_steps)):
+                if t == args.num_steps - 1:
+                    nextnonterminal = 1.0 - next_done
+                    nextvalues = next_value
+                else:
+                    nextnonterminal = 1.0 - dones[t + 1]
+                    nextvalues = values[t + 1]
+                delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
+                advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
+            returns = advantages + values
+
+        # flatten batch
+        b_obs = obs.reshape((-1,) + envs.single_observation_space.shape)
+        b_logprobs = logprobs.reshape(-1)
+        b_actions = actions.reshape((-1,) + envs.single_action_space.shape)
+        b_advantages = advantages.reshape(-1)
+        b_returns = returns.reshape(-1)
+        b_values = values.reshape(-1)
+
+        # update
+        b_inds = np.arange(args.batch_size)
+        clipfracs = []
+        for epoch in range(args.update_epochs):
+            np.random.shuffle(b_inds)
+            for start in range(0, args.batch_size, args.minibatch_size):
+                end = start + args.minibatch_size
+                mb_inds = b_inds[start:end]
+
+                _, newlogprob, entropy, newvalue = agent.get_action_and_value(b_obs[mb_inds], b_actions.long()[mb_inds])
+                logratio = newlogprob - b_logprobs[mb_inds]
+                ratio = logratio.exp()
+
+                with torch.no_grad():
+                    old_approx_kl = (-logratio).mean()
+                    approx_kl = ((ratio - 1) - logratio).mean()
+                    clipfracs += [((ratio - 1.0).abs() > args.clip_coef).float().mean().item()]
+
+                mb_advantages = b_advantages[mb_inds]
+                if args.norm_adv:
+                    mb_advantages = (mb_advantages - mb_advantages.mean()) / (mb_advantages.std() + 1e-8)
+
+                pg_loss1 = -mb_advantages * ratio
+                pg_loss2 = -mb_advantages * torch.clamp(ratio, 1 - args.clip_coef, 1 + args.clip_coef)
+                pg_loss = torch.max(pg_loss1, pg_loss2).mean()
+
+                newvalue = newvalue.view(-1)
+                if args.clip_vloss:
+                    v_loss_unclipped = (newvalue - b_returns[mb_inds]) ** 2
+                    v_clipped = b_values[mb_inds] + torch.clamp(
+                        newvalue - b_values[mb_inds], -args.clip_coef, args.clip_coef,
+                    )
+                    v_loss_clipped = (v_clipped - b_returns[mb_inds]) ** 2
+                    v_loss = 0.5 * torch.max(v_loss_unclipped, v_loss_clipped).mean()
+                else:
+                    v_loss = 0.5 * ((newvalue - b_returns[mb_inds]) ** 2).mean()
+
+                entropy_loss = entropy.mean()
+                loss = pg_loss - args.ent_coef * entropy_loss + v_loss * args.vf_coef
+
+                optimizer.zero_grad()
+                loss.backward()
+                nn.utils.clip_grad_norm_(agent.parameters(), args.max_grad_norm)
+                optimizer.step()
+
+            if args.target_kl is not None and approx_kl > args.target_kl:
+                break
+
+        # metrics
+        y_pred, y_true = b_values.cpu().numpy(), b_returns.cpu().numpy()
+        var_y = np.var(y_true)
+        explained_var = np.nan if var_y == 0 else 1 - np.var(y_true - y_pred) / var_y
+
+        sps = int(global_step / (time.time() - start_time))
+        train_success_rate = float(np.mean(iter_successes)) if len(iter_successes) else 0.0
+        print(f"Iter {iteration:4d}/{args.num_iterations} | SPS: {sps:5d} | R: {rewards.mean().item():6.3f}")
+        if args.track:
+            try:
+                import wandb
+                wandb.log({
+                    "global_step": int(global_step),
+                    "charts/progress": float(100.0 * iteration / max(1, args.num_iterations)),
+                    "train/value_loss": float(v_loss.item()),
+                    "train/policy_loss": float(pg_loss.item()),
+                    "losses/value_loss": float(v_loss.item()),
+                    "losses/policy_loss": float(pg_loss.item()),
+                    "train/entropy": float(entropy_loss.item()),
+                    "train/old_approx_kl": float(old_approx_kl.item()),
+                    "train/approx_kl": float(approx_kl.item()),
+                    "train/clipfrac": float(np.mean(clipfracs)) if len(clipfracs) else 0.0,
+                    "losses/explained_variance": float(explained_var),
+                    "charts/avg_reward": float(rewards.mean().item()),
+                    "charts/avg_value": float(values.mean().item()),
+                    "perf/SPS": int(sps),
+                    "charts/SPS": int(sps),
+                    "train/success_rate": train_success_rate,
+                    "charts/train_success_rate": train_success_rate,
+                    "train/learning_rate": float(optimizer.param_groups[0]["lr"]),
+                }, step=global_step)
+            except Exception:
+                pass
+
+        # periodic evaluation collection
+        if iteration % eval_every_iters == 0:
+            try:
+                eval_thunk = make_env(0, run_name, args.seed + 9999, args.scramble_depth, args.max_steps_env, False)
+                collect_eval_trajectories(agent, eval_thunk, n_episodes=args.eval_episodes, step_tag=global_step)
+                if args.track:
+                    try:
+                        import json as _json
+                        from pathlib import Path as _Path
+                        mpath = _Path(f"runs/{run_name}/trajectories/step_{global_step}/metrics.json")
+                        if mpath.exists():
+                            with mpath.open("r") as mf:
+                                metrics = _json.load(mf)
+                            wandb.log({
+                                "eval/success_rate": metrics.get("success_rate"),
+                                "eval/avg_return": metrics.get("avg_return"),
+                                "eval/std_return": metrics.get("std_return"),
+                                "eval/episodes": metrics.get("episodes"),
+                            }, step=global_step)
+                    except Exception:
+                        pass
+                print(f"Collected {args.eval_episodes} eval trajectories at global_step {global_step}")
+            except Exception as e:
+                print(f"Warning: eval trajectory collection failed at step {global_step}: {e}")
+
+    envs.close()

wandb/run-20251218_182745-ahjkdup0/files/config.yaml

@@ -0,0 +1,201 @@
+_wandb:
+    value:
+        cli_version: 0.23.0
+        code_path: code/cleanrl/cleanrl/ppo_rubikscube.py
+        e:
+            fcxjy4miz5x8bvnkek48fociofug1x4f:
+                args:
+                    - --track
+                codePath: cleanrl/cleanrl/ppo_rubikscube.py
+                codePathLocal: cleanrl/cleanrl/ppo_rubikscube.py
+                cpu_count: 64
+                cpu_count_logical: 128
+                cudaVersion: "12.4"
+                disk:
+                    /:
+                        total: "5153960755200"
+                        used: "447457550336"
+                email: haoyu-wa22@mails.tsinghua.edu.cn
+                executable: /root/local/miniconda3/envs/ragen/bin/python
+                git:
+                    commit: 8d73639b99b38265453f898b8d6af7d4af50d56e
+                    remote: https://github.com/mll-lab-nu/RAGEN.git
+                gpu: NVIDIA H100 80GB HBM3
+                gpu_count: 8
+                gpu_nvidia:
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-35e2d43d-4067-82ce-90d4-def9e389bf28
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-af4135e3-88f2-e9ac-518d-502c75a85429
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-d7fdeeba-fe9b-ec03-d9f7-6724fe4266b5
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-ccc4f668-3882-5a8e-2c07-c5cd08f6f666
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-7b73c0cf-d3d5-e10c-7176-a43be1e41001
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-81b58d94-5d1f-8ec2-f9d2-fd56172ed177
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-03e8bc66-3b44-6794-49fd-5392fbdda6d1
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-b7cf0ec6-7c29-1179-dceb-09565da51890
+                host: pt-4d654cf4576f4d23ad3d3919f12932fe-worker-0
+                memory:
+                    total: "2163642122240"
+                os: Linux-5.14.0-284.25.1.el9_2.x86_64-x86_64-with-glibc2.35
+                program: /mnt/general/wanghy/RAGEN/cleanrl/cleanrl/ppo_rubikscube.py
+                python: CPython 3.10.19
+                root: /mnt/general/wanghy/RAGEN
+                startedAt: "2025-12-18T10:27:45.627440Z"
+                writerId: fcxjy4miz5x8bvnkek48fociofug1x4f
+        m:
+            - "1": global_step
+              "6":
+                - 3
+              "7": []
+            - "2": train/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": rollout/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": eval/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": losses/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": charts/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": perf/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+        python_version: 3.10.19
+        t:
+            "1":
+                - 1
+                - 11
+                - 30
+                - 49
+                - 50
+                - 51
+                - 105
+            "2":
+                - 1
+                - 11
+                - 30
+                - 49
+                - 50
+                - 51
+                - 105
+            "3":
+                - 7
+                - 13
+                - 16
+                - 61
+            "4": 3.10.19
+            "5": 0.23.0
+            "6": 4.57.1
+            "12": 0.23.0
+            "13": linux-x86_64
+anneal_lr:
+    value: true
+batch_size:
+    value: 1024
+capture_video:
+    value: false
+clip_coef:
+    value: 0.2
+clip_vloss:
+    value: true
+cuda:
+    value: true
+ent_coef:
+    value: 0.01
+env_id:
+    value: RubiksCube2x2
+eval_episodes:
+    value: 4000
+eval_splits:
+    value: 2
+exp_name:
+    value: ppo_rubikscube
+gae_lambda:
+    value: 0.95
+gamma:
+    value: 0.99
+learning_rate:
+    value: 0.00025
+max_grad_norm:
+    value: 0.5
+max_steps_env:
+    value: 20
+minibatch_size:
+    value: 256
+norm_adv:
+    value: true
+num_envs:
+    value: 8
+num_iterations:
+    value: 976
+num_minibatches:
+    value: 4
+num_steps:
+    value: 128
+scramble_depth:
+    value: 3
+seed:
+    value: 1
+target_kl:
+    value: null
+torch_deterministic:
+    value: true
+total_timesteps:
+    value: 1000000
+track:
+    value: true
+update_epochs:
+    value: 4
+vf_coef:
+    value: 0.5
+wandb_entity:
+    value: null
+wandb_project_name:
+    value: cleanRL

wandb/run-20251218_182745-ahjkdup0/files/diff.patch

@@ -0,0 +1,1162 @@
+diff --git a/config/base.yaml b/config/base.yaml
+index a1659e4..9df6250 100644
+--- a/config/base.yaml
++++ b/config/base.yaml
+@@ -9,15 +9,16 @@ seed:
+   train: 10000
+   val: 123
+ 
+-micro_batch_size_per_gpu: 4
++micro_batch_size_per_gpu: 1
+ ppo_mini_batch_size: 32
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/Qwen/Qwen2.5-0.5B-Instruct
++# Qwen/Qwen2.5-0.5B-Instruct
+ enable_response_mask: True # Enabling response mask could improve stability of rollout/old_log_prob, as P(st|history) are no longer calculated in loss here. See https://docs.google.com/document/d/1bg7obeiKTExuHHBl5uOiSpec5uLDZ2Tgvxy6li5pHX4/edit?usp=sharing for more details.
+ grpo_advantage_length_weight: False # if you do not enable this and critic/advantage_estimator is GRPO, and the critic/advantages/mean is too low, then you can try enabling this to encourage reasoning and forbid collapse
+ 
+ lora:
+   rank: 0
+-  alpha: 64
++  alpha: 16
+   target_modules: all-linear
+ 
+ actor_rollout_ref:
+@@ -46,10 +47,10 @@ actor_rollout_ref:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu} # following micro_batch_size_per_gpu
+     tensor_model_parallel_size: 1
+-    max_model_len: 3600
++    max_model_len: 7200 #3600 why** 14400
+     prompt_length: 1 # useless. Just put it here
+     response_length: 400 # single-turn response length
+-    gpu_memory_utilization: 0.5
++    gpu_memory_utilization: 0.7
+     max_num_batched_tokens: 8192 # set only when enable_chunked_prefill is true
+     temperature: 1
+     rollout_filter_ratio: 0.25
+@@ -90,26 +91,28 @@ algorithm:
+     kl_coef: 0.000
+ 
+ trainer:
+-  project_name: ragen_latest
++  project_name: 
+   experiment_name: test
+   local_log_dir: "results/"
+-  save_freq: 100
++  save_freq: -1
+   total_training_steps: 200
+   validation_steps: 1 # validation instances = validation_steps * val_env_groups * group_size
+   val_before_train: True
+   n_gpus_per_node: 1
+   test_freq: 10
+   generations_to_log_to_wandb: 
++    train: 128
+     val: 20
+   logger: [ 'console', 'wandb' ]
+   max_actor_ckpt_to_keep: 1
+   max_critic_ckpt_to_keep: 1
++  default_local_dir: /mnt/general/wanghy/RAGEN/saves/
+ 
+ agent_proxy:
+   max_context_window: -1 # set a value > 0 to enable context window for long trajectory
+-  max_turn: 5
++  max_turn: 25 #25 why** 700
+   action_sep: "||"
+-  max_actions_per_turn: 2 # how many actions can be output at most in a single turn
++  max_actions_per_turn: 1 # how many actions can be output at most in a single turn
+   use_turn_scores: False # important to GAE when applying token-level rewards to token-level advantages. If False, will take the sum of scores as the reward for the last turn.
+   enable_think: True # False -> no think RL
+   reward_normalization:
+@@ -126,11 +129,11 @@ es_manager:
+       tags: ["CoordSokoban"]
+       n_groups: [8] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+   val:
+-    env_groups: 32
+-    group_size: 16 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
++    env_groups: 256
++    group_size: 1 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
+     env_configs:
+       tags: ["CoordSokoban"]
+-      n_groups: [32] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      n_groups: [256] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ ctx_manager:
+   generation: # go to vllm
+diff --git a/config/envs.yaml b/config/envs.yaml
+index 1002ac8..ba3ac6f 100644
+--- a/config/envs.yaml
++++ b/config/envs.yaml
+@@ -2,6 +2,7 @@ custom_envs:
+   SimpleSokoban:
+     env_type: sokoban
+     max_actions_per_traj: 10 # used in environment state manager to control the actual max actions executed per trajectory
++    # ORIGNAL env_instruction
+     env_instruction: |
+       You are solving the Sokoban puzzle. 
+       You are the player and you need to push all boxes to targets. 
+@@ -12,7 +13,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+ 
+   LargerSokoban:
+@@ -65,7 +66,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+       observation_format: "grid_coord"
+ 
+@@ -111,14 +112,14 @@ custom_envs:
+ 
+   FrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: "You are solving the FrozenLake puzzle. Forbid the whole and go to the target. You may move to the unintended direction due to the slippery ice. Example answer format: <think>To forbid the hole and go to the target, I should go left then go up.</think><answer>Left || Up</answer>"
+     max_tokens: 100
+     env_config: null
+ 
+   CoordFrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: |
+       You are solving the FrozenLake puzzle. The observation includes both a symbol grid and zero-indexed coordinates for the start, goal, player, and any holes.
+       Coordinates range from the top-left corner (0, 0) to the bottom-right corner (5, 5).
+@@ -195,4 +196,80 @@ custom_envs:
+     max_actions_per_traj: 30
+     env_instruction: "You are a Lean theorem prover. Given a Lean theorem statement, propose a sequence of tactics that completes the proof. Think step by step about which tactics to apply next. Provide tactics separated by '||', for example <answer>intro || simp || rfl</answer>."
+     max_tokens: 512
+-    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
+
++    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
++
++  
++  game_2048:
++    env_type: game_2048
++    max_actions_per_traj: 700
++    env_instruction: |
++      You are playing the 2048 game on a 4x4 grid. Merge equal tiles by sliding Up, Right, Down, or Left.
++      If a move is invalid (no tiles move), a small penalty is applied. Respond with a single action.
++      Example: <answer>Up</answer>
++    max_tokens: 8192
++    env_config: null
++
++  blackjack:
++    env_type: blackjack
++    max_actions_per_traj: 10
++    env_instruction: |
++      You are playing Blackjack against a dealer. The dealer must hit on 16 or less and stand on 17 or more.
++      Choose either Stick or Hit. Respond with a single action.
++      Example: <answer>Hit</answer>
++    max_tokens: 64
++    env_config: null
++
++  rubikscube:
++    env_type: rubikscube
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.
++      Available actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.
++      - Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).
++      - Modifiers: A letter alone means 90° clockwise (e.g., 'R'). A letter with prime (') means 90° counter-clockwise (e.g., "R'").
++      Respond with a sequence of actions separated by "||".
++      Example: <answer>U</answer>
++    max_tokens: 96
++    env_config:
++      scramble_depth: 1
++      max_steps: 20
++      render_mode: "text"
++  
++  SimpleSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 2x2 box contains the numbers 1-4 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format, for example: <answer>place 1 at row 2 col 3</answer> or <answer>1,2,3</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 4
++      difficulty: "easy"
++      render_format: "with_feedback"
++      show_conflicts: false
++      show_valid_numbers: false
++      max_steps: 20
++
++  MediumSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 30
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 3x3 box contains the numbers 1-9 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format: <answer>place 5 at row 2 col 3</answer> or <answer>2,3,5</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 9
++      difficulty: "medium"
++      render_format: "with_feedback"
++      show_conflicts: true
++      show_valid_numbers: true
++      max_steps: 81
++
+diff --git a/config/eval.yaml b/config/eval.yaml
+index e65f3a5..7775fd1 100644
+--- a/config/eval.yaml
++++ b/config/eval.yaml
+@@ -1,14 +1,15 @@
+ defaults:
+   - envs
+-
++enable_response_mask: True
+ system:
+-  CUDA_VISIBLE_DEVICES: "0"
++  CUDA_VISIBLE_DEVICES: "0,1,2,3,4,5,6,7"
+ 
+ seed:
+   train: 10000
+   val: 123
+ 
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/openai/gpt-oss-20b
++# /mnt/general/share/model/tyzhu/SPA-frozenlake-qwen2.5-1.5b-instruct
+ 
+ lora:
+   rank: 0
+@@ -24,9 +25,9 @@ actor_rollout_ref:
+   rollout:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu}
+-    tensor_model_parallel_size: 1
++    tensor_model_parallel_size: 8
+     dtype: bfloat16
+-    max_model_len: 3600
++    max_model_len: 7200
+     prompt_length: 1
+     response_length: 400
+     gpu_memory_utilization: 0.9
+@@ -35,6 +36,7 @@ actor_rollout_ref:
+     free_cache_engine: True
+     enable_chunked_prefill: False
+     disable_log_stats: False
++    do_sample: True
+     val_kwargs:
+       do_sample: True
+       temperature: 0.5
+@@ -44,9 +46,9 @@ actor_rollout_ref:
+ 
+ agent_proxy:
+   max_context_window: -1
+-  max_turn: 5
++  max_turn: 25
+   action_sep: "||"
+-  max_actions_per_turn: 2
++  max_actions_per_turn: 1
+   use_turn_scores: False
+   enable_think: True
+   reward_normalization:
+@@ -59,13 +61,13 @@ es_manager:
+     env_groups: 8
+     group_size: 16
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [8]
+   val:
+     env_groups: 32
+-    group_size: 16
++    group_size: 128
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [32]
+ 
+ ctx_manager:
+diff --git a/config/evaluate_api_llm.yaml b/config/evaluate_api_llm.yaml
+index b21c95e..e2e6761 100644
+--- a/config/evaluate_api_llm.yaml
++++ b/config/evaluate_api_llm.yaml
+@@ -1,8 +1,11 @@
++#export OPENAI_BASE_URL="https://api.ohmygpt.com/v1"
++#export OPENAI_API_KEY="sk-o4sMxBkN5BB100C4D4a3T3BlBkFJF7791CA39EA14ca98041"
++#python -m ragen.eval_api hydra.searchpath='[file://./verl/verl/trainer/config]'
+ defaults:
+   - base # this is a symbolic link to the verl/verl/trainer/config/ppo_trainer.yaml file
+ 
+ model_config:
+-  model_name: gpt-4o # should be registered in model_info
++  model_name: ark-deepseek-v3-250324 # should be registered in model_info
+   max_concurrency: 16
+ 
+ model_info:
+@@ -24,9 +27,9 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+-  gpt-4o:
++  gpt-4o-mini:
+     provider_name: openai
+-    model_name: gpt-4o
++    model_name: gpt-4o-mini
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+@@ -36,21 +39,42 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  ark-deepseek-v3-250324:
++    provider_name: openai
++    model_name: ark-deepseek-v3-250324
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
+   deepseek-v3:
+     provider_name: deepseek
+     model_name: deepseek-chat
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  glm-4.6:
++    provider_name: openai
++    model_name: glm-4.6
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
++  TA/openai/gpt-oss-120b:
++    provider_name: openai
++    model_name: TA/openai/gpt-oss-120b
++    generation_kwargs:
++      temperature: 0
++      max_tokens: 8192
++      # max_retries: 5
+ 
+-
+-
++agent_proxy:
++  max_turn: 5
+ es_manager:
+   val:
+-    env_groups: 256
++    env_groups: 128
+     group_size: 1 # should be set to 1 because val temperature is set to 0 and same prompt leads to same output
+     env_configs:
+-      tags: ["CoordSokoban"]
+-      n_groups: [256] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      tags: ["rubikscube"]
++      n_groups: [128] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ 
++rollout:
++  max_model_len: 7200
+
+Submodule external/webshop-minimal contains modified content
+diff --git a/external/webshop-minimal/requirements.txt b/external/webshop-minimal/requirements.txt
+index 5a1b04f..238ed5a 100644
+--- a/external/webshop-minimal/requirements.txt
++++ b/external/webshop-minimal/requirements.txt
+@@ -4,7 +4,7 @@ flask
+ html2text
+ rank_bm25
+ pyserini
+-faiss-cpu
++faiss-gpu
+ thefuzz
+ gdown
+ spacy
+diff --git a/ragen/env/__init__.py b/ragen/env/__init__.py
+index b0f3461..bc6838e 100644
+--- a/ragen/env/__init__.py
++++ b/ragen/env/__init__.py
+@@ -12,6 +12,14 @@ from .metamathqa.env import MetaMathQAEnv
+ from .metamathqa.config import MetaMathQAEnvConfig
+ from .lean.config import LeanEnvConfig
+ from .lean.env import LeanEnv
++from .game_2048.config import Game2048EnvConfig
++from .game_2048.env import Game2048Env
++from .blackjack.config import BlackjackEnvConfig
++from .blackjack.env import BlackjackEnv
++from .rubikscube.config import RubiksCube2x2Config
++from .rubikscube.env import RubiksCube2x2Env
++from .sudoku.config import SudokuEnvConfig
++from .sudoku.env import SudokuEnv
+ 
+ 
+ REGISTERED_ENVS = {
+@@ -22,6 +30,10 @@ REGISTERED_ENVS = {
+     # 'alfworld': AlfredTXTEnv,
+     'metamathqa': MetaMathQAEnv,
+     'lean': LeanEnv,
++    'game_2048': Game2048Env,
++    'blackjack': BlackjackEnv,
++    'rubikscube': RubiksCube2x2Env,
++    'sudoku': SudokuEnv,
+ }
+ 
+ REGISTERED_ENV_CONFIGS = {
+@@ -32,6 +44,10 @@ REGISTERED_ENV_CONFIGS = {
+     # 'alfworld': AlfredEnvConfig,
+     'metamathqa': MetaMathQAEnvConfig,
+     'lean': LeanEnvConfig,
++    'game_2048': Game2048EnvConfig,
++    'blackjack': BlackjackEnvConfig,
++    'rubikscube': RubiksCube2x2Config,
++    'sudoku': SudokuEnvConfig,
+ }
+ 
+ try:
+diff --git a/ragen/env/frozen_lake/config.py b/ragen/env/frozen_lake/config.py
+index de054f4..9950c34 100644
+--- a/ragen/env/frozen_lake/config.py
++++ b/ragen/env/frozen_lake/config.py
+@@ -8,7 +8,7 @@ class FrozenLakeEnvConfig:
+     size: int = 4
+     p: float = 0.9
+     success_rate: float = 0.8
+-    is_slippery: bool = True
++    is_slippery: bool = False
+     map_seed: Optional[int] = None
+     render_mode: str = "text"
+     observation_format: str = "grid"
+diff --git a/ragen/env/frozen_lake/env.py b/ragen/env/frozen_lake/env.py
+index 9ee3add..eef11d8 100644
+--- a/ragen/env/frozen_lake/env.py
++++ b/ragen/env/frozen_lake/env.py
+@@ -13,6 +13,7 @@ from ragen.env.base import BaseDiscreteActionEnv
+ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+     def __init__(self, config: FrozenLakeEnvConfig = FrozenLakeEnvConfig()):
+         # Using mappings directly from config
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.GRID_LOOKUP = config.grid_lookup
+         self.ACTION_LOOKUP = config.action_lookup
+@@ -95,6 +96,7 @@ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+ if __name__ == "__main__":
+     import matplotlib.pyplot as plt
+     config = FrozenLakeEnvConfig(size=4, is_slippery=True)
++    # import pdb;pdb.set_trace()
+     env = FrozenLakeEnv(config)
+     print(env.reset())
+     while True:
+diff --git a/ragen/env/sokoban/env.py b/ragen/env/sokoban/env.py
+index 17cd636..6922b54 100644
+--- a/ragen/env/sokoban/env.py
++++ b/ragen/env/sokoban/env.py
+@@ -33,6 +33,7 @@ class SokobanEnv(BaseDiscreteActionEnv, GymSokobanEnv):
+     def reset(self, seed=None, mode=None):
+         try:
+             with all_seed(seed):
++                # import pdb;pdb.set_trace()
+                 self.room_fixed, self.room_state, self.box_mapping, action_sequence = generate_room(
+                     dim=self.dim_room,
+                     num_steps=self.num_gen_steps,
+diff --git a/ragen/llm_agent/agent_proxy.py b/ragen/llm_agent/agent_proxy.py
+index c25b50d..f6c4088 100644
+--- a/ragen/llm_agent/agent_proxy.py
++++ b/ragen/llm_agent/agent_proxy.py
+@@ -287,6 +287,7 @@ def main(config):
+ 	proxy = LLMAgentProxy(config, actor_wg, tokenizer)
+ 	import time
+ 	start_time = time.time()
++	# import pdb;pdb.set_trace()
+ 	rollouts = proxy.rollout(
+ 		DataProto(
+ 			batch=None,
+diff --git a/ragen/llm_agent/base_llm.py b/ragen/llm_agent/base_llm.py
+index 358a2eb..b6f7797 100644
+--- a/ragen/llm_agent/base_llm.py
++++ b/ragen/llm_agent/base_llm.py
+@@ -33,6 +33,7 @@ class OpenAIProvider(LLMProvider):
+             raise ValueError("OpenAI API key not provided and not found in environment variables")
+         
+         self.client = AsyncOpenAI(api_key=self.api_key)
++        # import pdb;pdb.set_trace()
+     
+     async def generate(self, messages: List[Dict[str, str]], **kwargs) -> LLMResponse:
+         if "o1-mini" in self.model_name:
+@@ -46,10 +47,7 @@ class OpenAIProvider(LLMProvider):
+         )
+         if response.choices[0].finish_reason in ['length', 'content_filter']:
+             raise ValueError("Content filtered or length exceeded")
+-        return LLMResponse(
+-            content=response.choices[0].message.content,
+-            model_name=response.model
+-        )
++        return LLMResponse(content=response.choices[0].message.content,model_name=response.model)
+ 
+ class DeepSeekProvider(LLMProvider):
+     """DeepSeek API provider implementation"""
+@@ -199,7 +197,7 @@ class ConcurrentLLM:
+         
+         # Queue to store unfinished or failed tasks
+         current_batch = messages_list.copy()
+-        max_retries = kwargs.get("max_retries", 100)
++        max_retries = kwargs.get("max_retries", 10)
+         retry_count = 0
+         
+         while current_batch and retry_count < max_retries:
+diff --git a/ragen/llm_agent/ctx_manager.py b/ragen/llm_agent/ctx_manager.py
+index 905247a..af20a01 100644
+--- a/ragen/llm_agent/ctx_manager.py
++++ b/ragen/llm_agent/ctx_manager.py
+@@ -88,6 +88,7 @@ class ContextManager:
+         Initialize the ContextManager.
+         Processor is used to process the image data.
+         """
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.tokenizer = tokenizer
+         self.processor = processor
+@@ -311,7 +312,7 @@ class ContextManager:
+ 
+             llm_input_texts.append(text_with_prompt)
+             messages_list.append(messages)
+-
++        # import pdb;pdb.set_trace()
+         inputs = self.tokenizer(llm_input_texts, return_tensors="pt", padding=True, padding_side="left", truncation=False) # We have truncated previously, truncation in tokenizer may cause issues.
+         input_ids, attention_mask = inputs.input_ids, inputs.attention_mask
+         position_ids = (attention_mask.cumsum(dim=-1) - 1).clamp(min=0)
+@@ -353,6 +354,30 @@ class ContextManager:
+                 key: np.sum(value) / self.env_nums[key.split("/")[0]]
+                 for key, value in metrics.items()
+             }
++            # Derived metrics for wandb logging
++            try:
++                # charts/avg_episode_return: average across all 2048 env groups
++                ep_keys = [k for k in metrics.keys() if k.endswith('/episodic_return')]
++                if len(ep_keys) > 0:
++                    ep_vals = []
++                    for k in ep_keys:
++                        tag = k.split('/')[0]
++                        denom = self.env_nums.get(tag, max(1, len(metrics[k])))
++                        ep_vals.append(float(np.sum(metrics[k]) / denom))
++                    mean_metrics["charts/avg_episode_return"] = float(np.mean(ep_vals))
++            except Exception:
++                pass
++            try:
++                # rollout/max_tile: max over all envs in this batch
++                tile_keys = [k for k in metrics.keys() if k.endswith('/max_tile')]
++                if len(tile_keys) > 0:
++                    tile_vals = []
++                    for k in tile_keys:
++                        tile_vals.extend(metrics[k])
++                    if len(tile_vals) > 0:
++                        mean_metrics["rollout/max_tile"] = int(np.max(tile_vals))
++            except Exception:
++                pass
+             for key, values in metrics.items():
+                 if not isinstance(values, list):
+                     continue
+diff --git a/ragen/llm_agent/es_manager.py b/ragen/llm_agent/es_manager.py
+index b87a2b3..1a49d3e 100644
+--- a/ragen/llm_agent/es_manager.py
++++ b/ragen/llm_agent/es_manager.py
+@@ -128,18 +128,40 @@ class EnvStateManager:
+         env_outputs: List[Dict]
+             {env_id: int, history: List[Dict][{state: str, actions: List[str], reward: float, info: Dict, llm_response: str, llm_raw_response: str, (Optional)images: List[PIL.Image.Image]}]}
+         """
++        # def _execute_actions(env, actions):
++        #     acc_reward, turn_info, turn_done = 0, {}, False
++        #     executed_actions = []
++        #     for action in actions:
++        #         _, reward, done, info = env.step(action)
++        #         acc_reward += reward
++        #         turn_info.update(info) # NOTE: currently use last info for multi-action
++        #         executed_actions.append(action)
++        #         if done:
++        #             turn_done = True
++        #             break
++            
++        #     return acc_reward, turn_info, turn_done, executed_actions
+         def _execute_actions(env, actions):
+-            acc_reward, turn_info, turn_done = 0, {}, False
++            acc_reward, turn_info, turn_done = 0.0, {}, False
++            raw_acc_reward = 0.0
+             executed_actions = []
+             for action in actions:
+                 _, reward, done, info = env.step(action)
+-                acc_reward += reward
++                acc_reward += float(reward)
++                try:
++                    raw_acc_reward += float(info.get('raw_reward', 0.0))
++                except Exception:
++                    pass
+                 turn_info.update(info) # NOTE: currently use last info for multi-action
+                 executed_actions.append(action)
+                 if done:
+                     turn_done = True
+                     break
+-            
++            # Overwrite per-turn raw_reward to reflect the sum across all executed actions in this turn
++            try:
++                turn_info['raw_reward'] = float(raw_acc_reward)
++            except Exception:
++                pass
+             return acc_reward, turn_info, turn_done, executed_actions
+ 
+         def _log_env_state(status, history, cur_obs, max_actions_per_traj, executed_actions, all_actions, acc_reward, turn_done, turn_info, env_input):
+@@ -198,6 +220,18 @@ class EnvStateManager:
+                 'success': float(status.terminated and (not status.truncated)),
+                 'num_actions': status.num_actions,
+             }
++            # Add episodic-level metrics
++            # try:
++            #     # Sum of per-turn rewards equals the episodic return (env-shaped reward)
++            #     env_metric['episodic_return'] = float(sum(status.rewards))
++            # except Exception:
++            #     pass
++            try:
++                # Final max tile on the board at the end of the rollout
++                import numpy as _np
++                env_metric['max_tile'] = int(_np.max(entry['env'].grid))
++            except Exception:
++                pass
+             custom_metric = {}
+             for turn in cache['history']:
+                 for k, v in turn.get('info', {}).items():
+@@ -212,6 +246,12 @@ class EnvStateManager:
+                             "Skipping non-numeric metric '%s' with value %r for env %s.",
+                             k, v, entry['tag']
+                         )
++            # Add episodic_return as the SUM of raw_reward across steps (align with CleanRL)
++            try:
++                if 'raw_reward' in custom_metric:
++                    env_metric['episodic_return'] = float(np.sum(custom_metric['raw_reward']))
++            except Exception:
++                pass
+             for k, v in custom_metric.items():
+                 # TODO: Move TURN_LVL_METRICS into the environment
+                 if "webshop" not in cache['tag'].lower() or ("webshop" in cache['tag'].lower() and k in TURN_LVL_METRICS):
+@@ -219,7 +259,12 @@ class EnvStateManager:
+                 else:
+                     env_metric['traj_sum/' + k] = np.sum(v)
+ 
+-
++            try:
++                if 'score' in custom_metric and len(custom_metric['score']) > 0:
++                    env_metric['final_score'] = float(custom_metric['score'][-1])
++            except Exception:
++                pass
++                
+             cache['history'][-1]['metrics'] = custom_metric
+             env_metric = {f"{entry['tag']}/{k}": v for k, v in env_metric.items()}
+             cache['metrics'] = env_metric
+diff --git a/requirements.txt b/requirements.txt
+index 2fd756e..9b521bf 100644
+--- a/requirements.txt
++++ b/requirements.txt
+@@ -7,7 +7,6 @@ accelerate
+ codetiming
+ datasets
+ dill
+-flash-attn==2.7.4.post1
+ hydra-core
+ numpy
+ pandas
+@@ -15,19 +14,19 @@ pybind11
+ ray>=2.10
+ tensordict>=0.8.0,<0.9.0
+ transformers
+-vllm==0.8.2
++vllm==0.8.5
+ wandb
+ gymnasium
+ gymnasium[toy-text]
+ 
+ pyarrow>=15.0.0
+ pylatexenc
+-torchdata
++# torchdata
+ debugpy
+ 
+ together
+ anthropic
+-faiss-cpu==1.11.0
++faiss-gpu
+ 
+ # This is optional, but needs to be installed with main requirements if you want to use webshop
+ -r external/webshop-minimal/requirements.txt
+diff --git a/scripts/setup_ragen.sh b/scripts/setup_ragen.sh
+index f9a7cd9..85a93c4 100644
+--- a/scripts/setup_ragen.sh
++++ b/scripts/setup_ragen.sh
+@@ -94,10 +94,10 @@ main() {
+         pip install torch==2.5.0 --index-url https://download.pytorch.org/whl/cu124
+         
+         print_step "Installing flash-attention..."
+-        pip3 install flash-attn==2.7.4.post1 --no-build-isolation
++        # pip3 install flash-attn==2.7.4.post1 --no-build-isolation
+     else
+         print_step "Installing PyTorch without CUDA support..."
+-        pip install torch==2.5.0
++        pip install torch==2.4.0
+     fi
+     
+     # Install remaining requirements
+@@ -137,8 +137,8 @@ main() {
+     conda install conda-forge::gdown
+     mkdir -p external/webshop-minimal/webshop_minimal/data/full
+     cd external/webshop-minimal/webshop_minimal/data/full
+-    gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
+-    gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
++    # gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
++    # gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
+     cd ../../../../..
+ 
+     echo -e "${GREEN}Installation completed successfully!${NC}"
+diff --git a/train_all.sh b/train_all.sh
+index 0157306..33035f5 100755
+--- a/train_all.sh
++++ b/train_all.sh
+@@ -6,246 +6,257 @@ USE_GRPO="algorithm.adv_estimator=grpo"
+ USE_PPO="algorithm.adv_estimator=gae" # by default.
+ USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+ 
++
++# python train.py --config-name _8_2048 system.CUDA_VISIBLE_DEVICES="'0,1,2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=game_2048 $USE_PPO $USE_BASE 
++
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-1 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-1.5B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'4,5'" trainer.project_name=ragen_latest_qwen_25_3b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-3B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-3 $USE_PPO $USE_BASE 
++
+ # Section 3.1&3.2 - General Observations
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozen_lake-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozen_lake-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=bandit-ppo-multitask $USE_PPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'7'" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-ppo-frommlp $USE_PPO $USE_BASE 
+ 
+-# Section 4.1 - Filtering and critic
+-# 0.25
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-
+-wait
+-
+-# 0.5
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-
+-# 0.75
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-
+-wait
+-
+-# Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=sokoban-ppo-box1-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-wait
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.project_name=ragen_latest_qwen_25_3b_it  trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_3B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-# Section 5.1 - Reasoning Helps Generalization
++# # Section 4.1 - Filtering and critic
++# # 0.25
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
+-    agent_proxy.enable_think=False
++# # 0.5
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++
++# # 0.75
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
+-    agent_proxy.enable_think=False
++# # Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+ 
++# wait
+ 
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
++# # Section 5.1 - Reasoning Helps Generalization
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
+ 
+-# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
++#     agent_proxy.enable_think=False
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader"
+ 
+-# COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
++#     agent_proxy.enable_think=False
+ 
+-wait
+ 
+-# Section 5.2 - what leads to better reasoning?
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
++# # SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
++# # COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
+ 
++# wait
+ 
+-wait
++# # Section 5.2 - what leads to better reasoning?
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
+ 
+-# Section 6.1 varying action count
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
++# wait
+ 
+-# section 6.2 Varying prompt diversity
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
+ 
+-wait 
++# wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
++# wait
+ 
++# # Section 6.1 varying action count
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
+ 
++# # section 6.2 Varying prompt diversity
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
+ 
++# wait 
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
+ 
+-# Extension: Training 7B reasoning model
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
+ 
+-# trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
++# # Extension: Training 7B reasoning model
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
+ 
++# python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
+ 
++# # trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_PPO="algorithm.adv_estimator=gae" # by default.
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
+-# enable_response_mask: False
+-# grpo_advantage_length_weight: True
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+ 
+-# extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
++# # USE_PPO="algorithm.adv_estimator=gae" # by default.
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
+ 
+-# extension: bi-level gae
+-python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
+-    system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
+-    model_path=Qwen/Qwen2.5-0.5B-Instruct \
+-    algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
+-    agent_proxy.use_turn_scores=True \
+-    actor_rollout_ref.rollout.tp_size_check=False
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
+ 
+-# extension: webshop
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # enable_response_mask: False
++# # grpo_advantage_length_weight: True
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: bi-level gae
++# python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
++#     system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
++#     model_path=Qwen/Qwen2.5-0.5B-Instruct \
++#     algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
++#     agent_proxy.use_turn_scores=True \
++#     actor_rollout_ref.rollout.tp_size_check=False
+ 
+ 
+-# normal:sokoban
+-# extension: sokoban
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # extension: webshop
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
+ 
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++
++# # normal:sokoban
++# # extension: sokoban
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
++
++
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
++
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+ 
+ 
+-python train.py \
+-    trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+
++# python train.py \
++#     trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+

wandb/run-20251218_182745-ahjkdup0/files/diff_8d73639b99b38265453f898b8d6af7d4af50d56e.patch

@@ -0,0 +1,1162 @@
+diff --git a/config/base.yaml b/config/base.yaml
+index a1659e4..9df6250 100644
+--- a/config/base.yaml
++++ b/config/base.yaml
+@@ -9,15 +9,16 @@ seed:
+   train: 10000
+   val: 123
+ 
+-micro_batch_size_per_gpu: 4
++micro_batch_size_per_gpu: 1
+ ppo_mini_batch_size: 32
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/Qwen/Qwen2.5-0.5B-Instruct
++# Qwen/Qwen2.5-0.5B-Instruct
+ enable_response_mask: True # Enabling response mask could improve stability of rollout/old_log_prob, as P(st|history) are no longer calculated in loss here. See https://docs.google.com/document/d/1bg7obeiKTExuHHBl5uOiSpec5uLDZ2Tgvxy6li5pHX4/edit?usp=sharing for more details.
+ grpo_advantage_length_weight: False # if you do not enable this and critic/advantage_estimator is GRPO, and the critic/advantages/mean is too low, then you can try enabling this to encourage reasoning and forbid collapse
+ 
+ lora:
+   rank: 0
+-  alpha: 64
++  alpha: 16
+   target_modules: all-linear
+ 
+ actor_rollout_ref:
+@@ -46,10 +47,10 @@ actor_rollout_ref:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu} # following micro_batch_size_per_gpu
+     tensor_model_parallel_size: 1
+-    max_model_len: 3600
++    max_model_len: 7200 #3600 why** 14400
+     prompt_length: 1 # useless. Just put it here
+     response_length: 400 # single-turn response length
+-    gpu_memory_utilization: 0.5
++    gpu_memory_utilization: 0.7
+     max_num_batched_tokens: 8192 # set only when enable_chunked_prefill is true
+     temperature: 1
+     rollout_filter_ratio: 0.25
+@@ -90,26 +91,28 @@ algorithm:
+     kl_coef: 0.000
+ 
+ trainer:
+-  project_name: ragen_latest
++  project_name: 
+   experiment_name: test
+   local_log_dir: "results/"
+-  save_freq: 100
++  save_freq: -1
+   total_training_steps: 200
+   validation_steps: 1 # validation instances = validation_steps * val_env_groups * group_size
+   val_before_train: True
+   n_gpus_per_node: 1
+   test_freq: 10
+   generations_to_log_to_wandb: 
++    train: 128
+     val: 20
+   logger: [ 'console', 'wandb' ]
+   max_actor_ckpt_to_keep: 1
+   max_critic_ckpt_to_keep: 1
++  default_local_dir: /mnt/general/wanghy/RAGEN/saves/
+ 
+ agent_proxy:
+   max_context_window: -1 # set a value > 0 to enable context window for long trajectory
+-  max_turn: 5
++  max_turn: 25 #25 why** 700
+   action_sep: "||"
+-  max_actions_per_turn: 2 # how many actions can be output at most in a single turn
++  max_actions_per_turn: 1 # how many actions can be output at most in a single turn
+   use_turn_scores: False # important to GAE when applying token-level rewards to token-level advantages. If False, will take the sum of scores as the reward for the last turn.
+   enable_think: True # False -> no think RL
+   reward_normalization:
+@@ -126,11 +129,11 @@ es_manager:
+       tags: ["CoordSokoban"]
+       n_groups: [8] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+   val:
+-    env_groups: 32
+-    group_size: 16 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
++    env_groups: 256
++    group_size: 1 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
+     env_configs:
+       tags: ["CoordSokoban"]
+-      n_groups: [32] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      n_groups: [256] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ ctx_manager:
+   generation: # go to vllm
+diff --git a/config/envs.yaml b/config/envs.yaml
+index 1002ac8..ba3ac6f 100644
+--- a/config/envs.yaml
++++ b/config/envs.yaml
+@@ -2,6 +2,7 @@ custom_envs:
+   SimpleSokoban:
+     env_type: sokoban
+     max_actions_per_traj: 10 # used in environment state manager to control the actual max actions executed per trajectory
++    # ORIGNAL env_instruction
+     env_instruction: |
+       You are solving the Sokoban puzzle. 
+       You are the player and you need to push all boxes to targets. 
+@@ -12,7 +13,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+ 
+   LargerSokoban:
+@@ -65,7 +66,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+       observation_format: "grid_coord"
+ 
+@@ -111,14 +112,14 @@ custom_envs:
+ 
+   FrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: "You are solving the FrozenLake puzzle. Forbid the whole and go to the target. You may move to the unintended direction due to the slippery ice. Example answer format: <think>To forbid the hole and go to the target, I should go left then go up.</think><answer>Left || Up</answer>"
+     max_tokens: 100
+     env_config: null
+ 
+   CoordFrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: |
+       You are solving the FrozenLake puzzle. The observation includes both a symbol grid and zero-indexed coordinates for the start, goal, player, and any holes.
+       Coordinates range from the top-left corner (0, 0) to the bottom-right corner (5, 5).
+@@ -195,4 +196,80 @@ custom_envs:
+     max_actions_per_traj: 30
+     env_instruction: "You are a Lean theorem prover. Given a Lean theorem statement, propose a sequence of tactics that completes the proof. Think step by step about which tactics to apply next. Provide tactics separated by '||', for example <answer>intro || simp || rfl</answer>."
+     max_tokens: 512
+-    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
+
++    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
++
++  
++  game_2048:
++    env_type: game_2048
++    max_actions_per_traj: 700
++    env_instruction: |
++      You are playing the 2048 game on a 4x4 grid. Merge equal tiles by sliding Up, Right, Down, or Left.
++      If a move is invalid (no tiles move), a small penalty is applied. Respond with a single action.
++      Example: <answer>Up</answer>
++    max_tokens: 8192
++    env_config: null
++
++  blackjack:
++    env_type: blackjack
++    max_actions_per_traj: 10
++    env_instruction: |
++      You are playing Blackjack against a dealer. The dealer must hit on 16 or less and stand on 17 or more.
++      Choose either Stick or Hit. Respond with a single action.
++      Example: <answer>Hit</answer>
++    max_tokens: 64
++    env_config: null
++
++  rubikscube:
++    env_type: rubikscube
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.
++      Available actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.
++      - Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).
++      - Modifiers: A letter alone means 90° clockwise (e.g., 'R'). A letter with prime (') means 90° counter-clockwise (e.g., "R'").
++      Respond with a sequence of actions separated by "||".
++      Example: <answer>U</answer>
++    max_tokens: 96
++    env_config:
++      scramble_depth: 1
++      max_steps: 20
++      render_mode: "text"
++  
++  SimpleSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 2x2 box contains the numbers 1-4 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format, for example: <answer>place 1 at row 2 col 3</answer> or <answer>1,2,3</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 4
++      difficulty: "easy"
++      render_format: "with_feedback"
++      show_conflicts: false
++      show_valid_numbers: false
++      max_steps: 20
++
++  MediumSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 30
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 3x3 box contains the numbers 1-9 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format: <answer>place 5 at row 2 col 3</answer> or <answer>2,3,5</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 9
++      difficulty: "medium"
++      render_format: "with_feedback"
++      show_conflicts: true
++      show_valid_numbers: true
++      max_steps: 81
++
+diff --git a/config/eval.yaml b/config/eval.yaml
+index e65f3a5..7775fd1 100644
+--- a/config/eval.yaml
++++ b/config/eval.yaml
+@@ -1,14 +1,15 @@
+ defaults:
+   - envs
+-
++enable_response_mask: True
+ system:
+-  CUDA_VISIBLE_DEVICES: "0"
++  CUDA_VISIBLE_DEVICES: "0,1,2,3,4,5,6,7"
+ 
+ seed:
+   train: 10000
+   val: 123
+ 
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/openai/gpt-oss-20b
++# /mnt/general/share/model/tyzhu/SPA-frozenlake-qwen2.5-1.5b-instruct
+ 
+ lora:
+   rank: 0
+@@ -24,9 +25,9 @@ actor_rollout_ref:
+   rollout:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu}
+-    tensor_model_parallel_size: 1
++    tensor_model_parallel_size: 8
+     dtype: bfloat16
+-    max_model_len: 3600
++    max_model_len: 7200
+     prompt_length: 1
+     response_length: 400
+     gpu_memory_utilization: 0.9
+@@ -35,6 +36,7 @@ actor_rollout_ref:
+     free_cache_engine: True
+     enable_chunked_prefill: False
+     disable_log_stats: False
++    do_sample: True
+     val_kwargs:
+       do_sample: True
+       temperature: 0.5
+@@ -44,9 +46,9 @@ actor_rollout_ref:
+ 
+ agent_proxy:
+   max_context_window: -1
+-  max_turn: 5
++  max_turn: 25
+   action_sep: "||"
+-  max_actions_per_turn: 2
++  max_actions_per_turn: 1
+   use_turn_scores: False
+   enable_think: True
+   reward_normalization:
+@@ -59,13 +61,13 @@ es_manager:
+     env_groups: 8
+     group_size: 16
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [8]
+   val:
+     env_groups: 32
+-    group_size: 16
++    group_size: 128
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [32]
+ 
+ ctx_manager:
+diff --git a/config/evaluate_api_llm.yaml b/config/evaluate_api_llm.yaml
+index b21c95e..e2e6761 100644
+--- a/config/evaluate_api_llm.yaml
++++ b/config/evaluate_api_llm.yaml
+@@ -1,8 +1,11 @@
++#export OPENAI_BASE_URL="https://api.ohmygpt.com/v1"
++#export OPENAI_API_KEY="sk-o4sMxBkN5BB100C4D4a3T3BlBkFJF7791CA39EA14ca98041"
++#python -m ragen.eval_api hydra.searchpath='[file://./verl/verl/trainer/config]'
+ defaults:
+   - base # this is a symbolic link to the verl/verl/trainer/config/ppo_trainer.yaml file
+ 
+ model_config:
+-  model_name: gpt-4o # should be registered in model_info
++  model_name: ark-deepseek-v3-250324 # should be registered in model_info
+   max_concurrency: 16
+ 
+ model_info:
+@@ -24,9 +27,9 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+-  gpt-4o:
++  gpt-4o-mini:
+     provider_name: openai
+-    model_name: gpt-4o
++    model_name: gpt-4o-mini
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+@@ -36,21 +39,42 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  ark-deepseek-v3-250324:
++    provider_name: openai
++    model_name: ark-deepseek-v3-250324
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
+   deepseek-v3:
+     provider_name: deepseek
+     model_name: deepseek-chat
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  glm-4.6:
++    provider_name: openai
++    model_name: glm-4.6
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
++  TA/openai/gpt-oss-120b:
++    provider_name: openai
++    model_name: TA/openai/gpt-oss-120b
++    generation_kwargs:
++      temperature: 0
++      max_tokens: 8192
++      # max_retries: 5
+ 
+-
+-
++agent_proxy:
++  max_turn: 5
+ es_manager:
+   val:
+-    env_groups: 256
++    env_groups: 128
+     group_size: 1 # should be set to 1 because val temperature is set to 0 and same prompt leads to same output
+     env_configs:
+-      tags: ["CoordSokoban"]
+-      n_groups: [256] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      tags: ["rubikscube"]
++      n_groups: [128] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ 
++rollout:
++  max_model_len: 7200
+
+Submodule external/webshop-minimal contains modified content
+diff --git a/external/webshop-minimal/requirements.txt b/external/webshop-minimal/requirements.txt
+index 5a1b04f..238ed5a 100644
+--- a/external/webshop-minimal/requirements.txt
++++ b/external/webshop-minimal/requirements.txt
+@@ -4,7 +4,7 @@ flask
+ html2text
+ rank_bm25
+ pyserini
+-faiss-cpu
++faiss-gpu
+ thefuzz
+ gdown
+ spacy
+diff --git a/ragen/env/__init__.py b/ragen/env/__init__.py
+index b0f3461..bc6838e 100644
+--- a/ragen/env/__init__.py
++++ b/ragen/env/__init__.py
+@@ -12,6 +12,14 @@ from .metamathqa.env import MetaMathQAEnv
+ from .metamathqa.config import MetaMathQAEnvConfig
+ from .lean.config import LeanEnvConfig
+ from .lean.env import LeanEnv
++from .game_2048.config import Game2048EnvConfig
++from .game_2048.env import Game2048Env
++from .blackjack.config import BlackjackEnvConfig
++from .blackjack.env import BlackjackEnv
++from .rubikscube.config import RubiksCube2x2Config
++from .rubikscube.env import RubiksCube2x2Env
++from .sudoku.config import SudokuEnvConfig
++from .sudoku.env import SudokuEnv
+ 
+ 
+ REGISTERED_ENVS = {
+@@ -22,6 +30,10 @@ REGISTERED_ENVS = {
+     # 'alfworld': AlfredTXTEnv,
+     'metamathqa': MetaMathQAEnv,
+     'lean': LeanEnv,
++    'game_2048': Game2048Env,
++    'blackjack': BlackjackEnv,
++    'rubikscube': RubiksCube2x2Env,
++    'sudoku': SudokuEnv,
+ }
+ 
+ REGISTERED_ENV_CONFIGS = {
+@@ -32,6 +44,10 @@ REGISTERED_ENV_CONFIGS = {
+     # 'alfworld': AlfredEnvConfig,
+     'metamathqa': MetaMathQAEnvConfig,
+     'lean': LeanEnvConfig,
++    'game_2048': Game2048EnvConfig,
++    'blackjack': BlackjackEnvConfig,
++    'rubikscube': RubiksCube2x2Config,
++    'sudoku': SudokuEnvConfig,
+ }
+ 
+ try:
+diff --git a/ragen/env/frozen_lake/config.py b/ragen/env/frozen_lake/config.py
+index de054f4..9950c34 100644
+--- a/ragen/env/frozen_lake/config.py
++++ b/ragen/env/frozen_lake/config.py
+@@ -8,7 +8,7 @@ class FrozenLakeEnvConfig:
+     size: int = 4
+     p: float = 0.9
+     success_rate: float = 0.8
+-    is_slippery: bool = True
++    is_slippery: bool = False
+     map_seed: Optional[int] = None
+     render_mode: str = "text"
+     observation_format: str = "grid"
+diff --git a/ragen/env/frozen_lake/env.py b/ragen/env/frozen_lake/env.py
+index 9ee3add..eef11d8 100644
+--- a/ragen/env/frozen_lake/env.py
++++ b/ragen/env/frozen_lake/env.py
+@@ -13,6 +13,7 @@ from ragen.env.base import BaseDiscreteActionEnv
+ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+     def __init__(self, config: FrozenLakeEnvConfig = FrozenLakeEnvConfig()):
+         # Using mappings directly from config
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.GRID_LOOKUP = config.grid_lookup
+         self.ACTION_LOOKUP = config.action_lookup
+@@ -95,6 +96,7 @@ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+ if __name__ == "__main__":
+     import matplotlib.pyplot as plt
+     config = FrozenLakeEnvConfig(size=4, is_slippery=True)
++    # import pdb;pdb.set_trace()
+     env = FrozenLakeEnv(config)
+     print(env.reset())
+     while True:
+diff --git a/ragen/env/sokoban/env.py b/ragen/env/sokoban/env.py
+index 17cd636..6922b54 100644
+--- a/ragen/env/sokoban/env.py
++++ b/ragen/env/sokoban/env.py
+@@ -33,6 +33,7 @@ class SokobanEnv(BaseDiscreteActionEnv, GymSokobanEnv):
+     def reset(self, seed=None, mode=None):
+         try:
+             with all_seed(seed):
++                # import pdb;pdb.set_trace()
+                 self.room_fixed, self.room_state, self.box_mapping, action_sequence = generate_room(
+                     dim=self.dim_room,
+                     num_steps=self.num_gen_steps,
+diff --git a/ragen/llm_agent/agent_proxy.py b/ragen/llm_agent/agent_proxy.py
+index c25b50d..f6c4088 100644
+--- a/ragen/llm_agent/agent_proxy.py
++++ b/ragen/llm_agent/agent_proxy.py
+@@ -287,6 +287,7 @@ def main(config):
+ 	proxy = LLMAgentProxy(config, actor_wg, tokenizer)
+ 	import time
+ 	start_time = time.time()
++	# import pdb;pdb.set_trace()
+ 	rollouts = proxy.rollout(
+ 		DataProto(
+ 			batch=None,
+diff --git a/ragen/llm_agent/base_llm.py b/ragen/llm_agent/base_llm.py
+index 358a2eb..b6f7797 100644
+--- a/ragen/llm_agent/base_llm.py
++++ b/ragen/llm_agent/base_llm.py
+@@ -33,6 +33,7 @@ class OpenAIProvider(LLMProvider):
+             raise ValueError("OpenAI API key not provided and not found in environment variables")
+         
+         self.client = AsyncOpenAI(api_key=self.api_key)
++        # import pdb;pdb.set_trace()
+     
+     async def generate(self, messages: List[Dict[str, str]], **kwargs) -> LLMResponse:
+         if "o1-mini" in self.model_name:
+@@ -46,10 +47,7 @@ class OpenAIProvider(LLMProvider):
+         )
+         if response.choices[0].finish_reason in ['length', 'content_filter']:
+             raise ValueError("Content filtered or length exceeded")
+-        return LLMResponse(
+-            content=response.choices[0].message.content,
+-            model_name=response.model
+-        )
++        return LLMResponse(content=response.choices[0].message.content,model_name=response.model)
+ 
+ class DeepSeekProvider(LLMProvider):
+     """DeepSeek API provider implementation"""
+@@ -199,7 +197,7 @@ class ConcurrentLLM:
+         
+         # Queue to store unfinished or failed tasks
+         current_batch = messages_list.copy()
+-        max_retries = kwargs.get("max_retries", 100)
++        max_retries = kwargs.get("max_retries", 10)
+         retry_count = 0
+         
+         while current_batch and retry_count < max_retries:
+diff --git a/ragen/llm_agent/ctx_manager.py b/ragen/llm_agent/ctx_manager.py
+index 905247a..af20a01 100644
+--- a/ragen/llm_agent/ctx_manager.py
++++ b/ragen/llm_agent/ctx_manager.py
+@@ -88,6 +88,7 @@ class ContextManager:
+         Initialize the ContextManager.
+         Processor is used to process the image data.
+         """
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.tokenizer = tokenizer
+         self.processor = processor
+@@ -311,7 +312,7 @@ class ContextManager:
+ 
+             llm_input_texts.append(text_with_prompt)
+             messages_list.append(messages)
+-
++        # import pdb;pdb.set_trace()
+         inputs = self.tokenizer(llm_input_texts, return_tensors="pt", padding=True, padding_side="left", truncation=False) # We have truncated previously, truncation in tokenizer may cause issues.
+         input_ids, attention_mask = inputs.input_ids, inputs.attention_mask
+         position_ids = (attention_mask.cumsum(dim=-1) - 1).clamp(min=0)
+@@ -353,6 +354,30 @@ class ContextManager:
+                 key: np.sum(value) / self.env_nums[key.split("/")[0]]
+                 for key, value in metrics.items()
+             }
++            # Derived metrics for wandb logging
++            try:
++                # charts/avg_episode_return: average across all 2048 env groups
++                ep_keys = [k for k in metrics.keys() if k.endswith('/episodic_return')]
++                if len(ep_keys) > 0:
++                    ep_vals = []
++                    for k in ep_keys:
++                        tag = k.split('/')[0]
++                        denom = self.env_nums.get(tag, max(1, len(metrics[k])))
++                        ep_vals.append(float(np.sum(metrics[k]) / denom))
++                    mean_metrics["charts/avg_episode_return"] = float(np.mean(ep_vals))
++            except Exception:
++                pass
++            try:
++                # rollout/max_tile: max over all envs in this batch
++                tile_keys = [k for k in metrics.keys() if k.endswith('/max_tile')]
++                if len(tile_keys) > 0:
++                    tile_vals = []
++                    for k in tile_keys:
++                        tile_vals.extend(metrics[k])
++                    if len(tile_vals) > 0:
++                        mean_metrics["rollout/max_tile"] = int(np.max(tile_vals))
++            except Exception:
++                pass
+             for key, values in metrics.items():
+                 if not isinstance(values, list):
+                     continue
+diff --git a/ragen/llm_agent/es_manager.py b/ragen/llm_agent/es_manager.py
+index b87a2b3..1a49d3e 100644
+--- a/ragen/llm_agent/es_manager.py
++++ b/ragen/llm_agent/es_manager.py
+@@ -128,18 +128,40 @@ class EnvStateManager:
+         env_outputs: List[Dict]
+             {env_id: int, history: List[Dict][{state: str, actions: List[str], reward: float, info: Dict, llm_response: str, llm_raw_response: str, (Optional)images: List[PIL.Image.Image]}]}
+         """
++        # def _execute_actions(env, actions):
++        #     acc_reward, turn_info, turn_done = 0, {}, False
++        #     executed_actions = []
++        #     for action in actions:
++        #         _, reward, done, info = env.step(action)
++        #         acc_reward += reward
++        #         turn_info.update(info) # NOTE: currently use last info for multi-action
++        #         executed_actions.append(action)
++        #         if done:
++        #             turn_done = True
++        #             break
++            
++        #     return acc_reward, turn_info, turn_done, executed_actions
+         def _execute_actions(env, actions):
+-            acc_reward, turn_info, turn_done = 0, {}, False
++            acc_reward, turn_info, turn_done = 0.0, {}, False
++            raw_acc_reward = 0.0
+             executed_actions = []
+             for action in actions:
+                 _, reward, done, info = env.step(action)
+-                acc_reward += reward
++                acc_reward += float(reward)
++                try:
++                    raw_acc_reward += float(info.get('raw_reward', 0.0))
++                except Exception:
++                    pass
+                 turn_info.update(info) # NOTE: currently use last info for multi-action
+                 executed_actions.append(action)
+                 if done:
+                     turn_done = True
+                     break
+-            
++            # Overwrite per-turn raw_reward to reflect the sum across all executed actions in this turn
++            try:
++                turn_info['raw_reward'] = float(raw_acc_reward)
++            except Exception:
++                pass
+             return acc_reward, turn_info, turn_done, executed_actions
+ 
+         def _log_env_state(status, history, cur_obs, max_actions_per_traj, executed_actions, all_actions, acc_reward, turn_done, turn_info, env_input):
+@@ -198,6 +220,18 @@ class EnvStateManager:
+                 'success': float(status.terminated and (not status.truncated)),
+                 'num_actions': status.num_actions,
+             }
++            # Add episodic-level metrics
++            # try:
++            #     # Sum of per-turn rewards equals the episodic return (env-shaped reward)
++            #     env_metric['episodic_return'] = float(sum(status.rewards))
++            # except Exception:
++            #     pass
++            try:
++                # Final max tile on the board at the end of the rollout
++                import numpy as _np
++                env_metric['max_tile'] = int(_np.max(entry['env'].grid))
++            except Exception:
++                pass
+             custom_metric = {}
+             for turn in cache['history']:
+                 for k, v in turn.get('info', {}).items():
+@@ -212,6 +246,12 @@ class EnvStateManager:
+                             "Skipping non-numeric metric '%s' with value %r for env %s.",
+                             k, v, entry['tag']
+                         )
++            # Add episodic_return as the SUM of raw_reward across steps (align with CleanRL)
++            try:
++                if 'raw_reward' in custom_metric:
++                    env_metric['episodic_return'] = float(np.sum(custom_metric['raw_reward']))
++            except Exception:
++                pass
+             for k, v in custom_metric.items():
+                 # TODO: Move TURN_LVL_METRICS into the environment
+                 if "webshop" not in cache['tag'].lower() or ("webshop" in cache['tag'].lower() and k in TURN_LVL_METRICS):
+@@ -219,7 +259,12 @@ class EnvStateManager:
+                 else:
+                     env_metric['traj_sum/' + k] = np.sum(v)
+ 
+-
++            try:
++                if 'score' in custom_metric and len(custom_metric['score']) > 0:
++                    env_metric['final_score'] = float(custom_metric['score'][-1])
++            except Exception:
++                pass
++                
+             cache['history'][-1]['metrics'] = custom_metric
+             env_metric = {f"{entry['tag']}/{k}": v for k, v in env_metric.items()}
+             cache['metrics'] = env_metric
+diff --git a/requirements.txt b/requirements.txt
+index 2fd756e..9b521bf 100644
+--- a/requirements.txt
++++ b/requirements.txt
+@@ -7,7 +7,6 @@ accelerate
+ codetiming
+ datasets
+ dill
+-flash-attn==2.7.4.post1
+ hydra-core
+ numpy
+ pandas
+@@ -15,19 +14,19 @@ pybind11
+ ray>=2.10
+ tensordict>=0.8.0,<0.9.0
+ transformers
+-vllm==0.8.2
++vllm==0.8.5
+ wandb
+ gymnasium
+ gymnasium[toy-text]
+ 
+ pyarrow>=15.0.0
+ pylatexenc
+-torchdata
++# torchdata
+ debugpy
+ 
+ together
+ anthropic
+-faiss-cpu==1.11.0
++faiss-gpu
+ 
+ # This is optional, but needs to be installed with main requirements if you want to use webshop
+ -r external/webshop-minimal/requirements.txt
+diff --git a/scripts/setup_ragen.sh b/scripts/setup_ragen.sh
+index f9a7cd9..85a93c4 100644
+--- a/scripts/setup_ragen.sh
++++ b/scripts/setup_ragen.sh
+@@ -94,10 +94,10 @@ main() {
+         pip install torch==2.5.0 --index-url https://download.pytorch.org/whl/cu124
+         
+         print_step "Installing flash-attention..."
+-        pip3 install flash-attn==2.7.4.post1 --no-build-isolation
++        # pip3 install flash-attn==2.7.4.post1 --no-build-isolation
+     else
+         print_step "Installing PyTorch without CUDA support..."
+-        pip install torch==2.5.0
++        pip install torch==2.4.0
+     fi
+     
+     # Install remaining requirements
+@@ -137,8 +137,8 @@ main() {
+     conda install conda-forge::gdown
+     mkdir -p external/webshop-minimal/webshop_minimal/data/full
+     cd external/webshop-minimal/webshop_minimal/data/full
+-    gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
+-    gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
++    # gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
++    # gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
+     cd ../../../../..
+ 
+     echo -e "${GREEN}Installation completed successfully!${NC}"
+diff --git a/train_all.sh b/train_all.sh
+index 0157306..33035f5 100755
+--- a/train_all.sh
++++ b/train_all.sh
+@@ -6,246 +6,257 @@ USE_GRPO="algorithm.adv_estimator=grpo"
+ USE_PPO="algorithm.adv_estimator=gae" # by default.
+ USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+ 
++
++# python train.py --config-name _8_2048 system.CUDA_VISIBLE_DEVICES="'0,1,2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=game_2048 $USE_PPO $USE_BASE 
++
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-1 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-1.5B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'4,5'" trainer.project_name=ragen_latest_qwen_25_3b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-3B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-3 $USE_PPO $USE_BASE 
++
+ # Section 3.1&3.2 - General Observations
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozen_lake-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozen_lake-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=bandit-ppo-multitask $USE_PPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'7'" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-ppo-frommlp $USE_PPO $USE_BASE 
+ 
+-# Section 4.1 - Filtering and critic
+-# 0.25
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-
+-wait
+-
+-# 0.5
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-
+-# 0.75
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-
+-wait
+-
+-# Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=sokoban-ppo-box1-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-wait
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.project_name=ragen_latest_qwen_25_3b_it  trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_3B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-# Section 5.1 - Reasoning Helps Generalization
++# # Section 4.1 - Filtering and critic
++# # 0.25
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
+-    agent_proxy.enable_think=False
++# # 0.5
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++
++# # 0.75
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
+-    agent_proxy.enable_think=False
++# # Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+ 
++# wait
+ 
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
++# # Section 5.1 - Reasoning Helps Generalization
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
+ 
+-# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
++#     agent_proxy.enable_think=False
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader"
+ 
+-# COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
++#     agent_proxy.enable_think=False
+ 
+-wait
+ 
+-# Section 5.2 - what leads to better reasoning?
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
++# # SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
++# # COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
+ 
++# wait
+ 
+-wait
++# # Section 5.2 - what leads to better reasoning?
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
+ 
+-# Section 6.1 varying action count
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
++# wait
+ 
+-# section 6.2 Varying prompt diversity
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
+ 
+-wait 
++# wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
++# wait
+ 
++# # Section 6.1 varying action count
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
+ 
++# # section 6.2 Varying prompt diversity
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
+ 
++# wait 
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
+ 
+-# Extension: Training 7B reasoning model
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
+ 
+-# trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
++# # Extension: Training 7B reasoning model
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
+ 
++# python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
+ 
++# # trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_PPO="algorithm.adv_estimator=gae" # by default.
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
+-# enable_response_mask: False
+-# grpo_advantage_length_weight: True
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+ 
+-# extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
++# # USE_PPO="algorithm.adv_estimator=gae" # by default.
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
+ 
+-# extension: bi-level gae
+-python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
+-    system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
+-    model_path=Qwen/Qwen2.5-0.5B-Instruct \
+-    algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
+-    agent_proxy.use_turn_scores=True \
+-    actor_rollout_ref.rollout.tp_size_check=False
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
+ 
+-# extension: webshop
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # enable_response_mask: False
++# # grpo_advantage_length_weight: True
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: bi-level gae
++# python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
++#     system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
++#     model_path=Qwen/Qwen2.5-0.5B-Instruct \
++#     algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
++#     agent_proxy.use_turn_scores=True \
++#     actor_rollout_ref.rollout.tp_size_check=False
+ 
+ 
+-# normal:sokoban
+-# extension: sokoban
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # extension: webshop
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
+ 
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++
++# # normal:sokoban
++# # extension: sokoban
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
++
++
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
++
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+ 
+ 
+-python train.py \
+-    trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+
++# python train.py \
++#     trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+

wandb/run-20251218_182745-ahjkdup0/files/requirements.txt

@@ -0,0 +1,315 @@
+colorama==0.4.6
+psutil==7.1.2
+ragen==0.1
+ragen==0.1
+setuptools==80.9.0
+wheel==0.45.1
+pip==25.3
+ragen==0.1
+pytz==2025.2
+nvidia-cusparselt-cu12==0.6.2
+mpmath==1.3.0
+typing_extensions==4.15.0
+sympy==1.13.1
+pylatexenc==2.10
+nvidia-cusolver-cu12==11.6.1.9
+numpy==1.26.4
+Markdown==3.10
+absl-py==2.3.1
+pyasn1_modules==0.4.2
+Werkzeug==3.1.3
+networkx==3.3
+MarkupSafe==2.1.5
+fsspec==2025.9.0
+filelock==3.19.1
+nvidia-cudnn-cu12==9.1.0.70
+Jinja2==3.1.6
+tensorboard==2.20.0
+triton==3.2.0
+pillow==11.3.0
+xxhash==3.6.0
+nvidia-nvtx-cu12==12.4.127
+nvidia-nvjitlink-cu12==12.4.127
+nvidia-nccl-cu12==2.21.5
+nvidia-curand-cu12==10.3.5.147
+wrapt==2.0.1
+nvidia-cufft-cu12==11.2.1.3
+pyjnius==1.7.0
+nvidia-cuda-runtime-cu12==12.4.127
+nvidia-cuda-nvrtc-cu12==12.4.127
+websockets==15.0.1
+nvidia-cuda-cupti-cu12==12.4.127
+zipp==3.23.0
+nvidia-cublas-cu12==12.4.5.8
+nvidia-cusparse-cu12==12.3.1.170
+flash_attn==2.8.3
+rsa==4.9.1
+einops==0.8.1
+py-cpuinfo==9.0.0
+pure_eval==0.2.3
+ptyprocess==0.7.0
+gym-notices==0.1.0
+flatbuffers==25.9.23
+fastrlock==0.8.3
+Farama-Notifications==0.0.4
+faiss-gpu==1.7.2
+antlr4-python3-runtime==4.9.3
+wcwidth==0.2.14
+wasabi==1.1.3
+uvloop==0.22.1
+urllib3==2.5.0
+tzdata==2025.2
+typing-inspection==0.4.2
+traitlets==5.14.3
+tqdm==4.67.1
+tomli==2.3.0
+threadpoolctl==3.6.0
+tenacity==9.1.2
+tabulate==0.9.0
+spacy-loggers==1.0.5
+spacy-legacy==3.0.12
+soupsieve==2.8
+sniffio==1.3.1
+smmap==5.0.2
+six==1.17.0
+shellingham==1.5.4
+sentencepiece==0.2.1
+scipy==1.15.3
+safetensors==0.6.2
+rpds-py==0.29.0
+rignore==0.7.6
+regex==2025.11.3
+RapidFuzz==3.14.3
+rank-bm25==0.2.2
+pyzmq==27.1.0
+PyYAML==6.0.3
+pytokens==0.3.0
+python-multipart==0.0.20
+python-json-logger==4.0.0
+python-dotenv==1.2.1
+PySocks==1.7.1
+pyparsing==3.2.5
+PyJWT==2.10.1
+Pygments==2.19.2
+pygame==2.6.1
+pydantic_core==2.41.5
+pycparser==2.23
+pycountry==24.6.1
+pybind11==3.0.1
+pyarrow==22.0.0
+psutil==7.1.3
+protobuf==4.25.8
+propcache==0.4.1
+prometheus_client==0.23.1
+platformdirs==4.5.0
+pexpect==4.9.0
+pathspec==0.12.1
+partial-json-parser==0.2.1.1.post6
+parso==0.8.5
+packaging==25.0
+orjson==3.11.4
+opentelemetry-semantic-conventions-ai==0.4.13
+opencv-python-headless==4.12.0.88
+ninja==1.13.0
+nest-asyncio==1.6.0
+mypy_extensions==1.1.0
+murmurhash==1.0.15
+multidict==6.7.0
+msgspec==0.19.0
+msgpack==1.1.2
+mdurl==0.1.2
+loguru==0.7.3
+llvmlite==0.44.0
+llguidance==0.7.30
+lark==1.2.2
+kiwisolver==1.4.9
+joblib==1.5.2
+jiter==0.12.0
+itsdangerous==2.2.0
+interegular==0.3.3
+ImageIO==2.37.2
+idna==3.11
+humanfriendly==10.0
+httpx-sse==0.4.3
+httptools==0.7.1
+html2text==2025.4.15
+hf-xet==1.2.0
+h11==0.16.0
+grpcio==1.76.0
+frozenlist==1.8.0
+fonttools==4.60.1
+executing==2.2.1
+exceptiongroup==1.3.0
+eval_type_backport==0.2.2
+docstring_parser==0.17.0
+dnspython==2.8.0
+distro==1.9.0
+diskcache==5.6.3
+dill==0.4.0
+decorator==5.2.1
+debugpy==1.8.17
+Cython==3.2.1
+cymem==2.0.13
+cycler==0.12.1
+cupy-cuda12x==13.6.0
+contourpy==1.3.2
+colorama==0.4.6
+codetiming==1.4.0
+cloudpickle==3.1.2
+cloudpathlib==0.23.0
+click==8.3.1
+charset-normalizer==3.4.4
+certifi==2025.11.12
+catalogue==2.0.10
+cachetools==6.2.2
+blis==1.3.2
+blinker==1.9.0
+blake3==1.0.8
+attrs==25.4.0
+async-timeout==5.0.1
+asttokens==3.0.1
+astor==0.8.1
+annotated-types==0.7.0
+annotated-doc==0.0.4
+airportsdata==20250909
+aiohappyeyeballs==2.6.1
+yarl==1.22.0
+uvicorn==0.38.0
+typer-slim==0.20.0
+thefuzz==0.22.1
+stack-data==0.6.3
+srsly==2.5.1
+smart_open==7.5.0
+sentry-sdk==2.44.0
+scikit-learn==1.7.2
+requests==2.32.5
+referencing==0.37.0
+python-dateutil==2.9.0.post0
+pydantic==2.12.4
+prompt_toolkit==3.0.52
+preshed==3.0.11
+opentelemetry-proto==1.26.0
+omegaconf==2.3.0
+numba==0.61.2
+nltk==3.9.2
+multiprocess==0.70.18
+matplotlib-inline==0.2.1
+markdown-it-py==4.0.0
+jedi==0.19.2
+importlib_metadata==8.0.0
+httpcore==1.0.9
+gymnasium==1.2.2
+gym==0.26.2
+googleapis-common-protos==1.72.0
+gitdb==4.0.12
+gguf==0.17.1
+Flask==3.1.2
+faiss-cpu==1.13.0
+email-validator==2.3.0
+depyf==0.18.0
+Deprecated==1.3.1
+coloredlogs==15.0.1
+cffi==2.0.0
+black==25.11.0
+beautifulsoup4==4.14.2
+anyio==4.11.0
+aiosignal==1.4.0
+watchfiles==1.1.1
+tiktoken==0.12.0
+starlette==0.49.3
+sse-starlette==3.0.3
+rich==14.2.0
+pydantic-settings==2.12.0
+pydantic-extra-types==2.10.6
+pandas==2.3.3
+opentelemetry-exporter-otlp-proto-common==1.26.0
+onnxruntime==1.23.2
+matplotlib==3.10.7
+lm-format-enforcer==0.10.12
+jsonschema-specifications==2025.9.1
+ipython==8.37.0
+hydra-core==1.3.2
+huggingface-hub==0.36.0
+httpx==0.28.1
+gym-sokoban==0.0.6
+GitPython==3.1.45
+cryptography==46.0.3
+confection==0.1.5
+cleantext==1.1.4
+aiohttp==3.13.2
+weasel==0.4.3
+wandb==0.23.0
+typer==0.19.2
+virtualenv==20.35.4
+pyasn1==0.6.1
+torch==2.6.0
+tokenizers==0.22.1
+thinc==8.3.9
+rich-toolkit==0.15.1
+prometheus-fastapi-instrumentator==7.1.0
+openai==2.8.0
+jsonschema==4.25.1
+gdown==5.2.0
+fastapi==0.121.2
+anthropic==0.73.0
+xformers==0.0.29.post2
+transformers==4.57.1
+proto-plus==1.26.1
+torchvision==0.21.0
+torchdata==0.11.0
+torchaudio==2.6.0
+together==1.5.30
+tensordict==0.8.3
+spacy==3.8.9
+ray==2.51.1
+outlines_core==0.1.26
+mistral_common==1.8.5
+mcp==1.21.1
+fastapi-cloud-cli==0.3.1
+fastapi-cli==0.0.16
+datasets==4.4.1
+accelerate==1.11.0
+xgrammar==0.1.18
+pyserini==1.2.0
+peft==0.18.0
+outlines==0.1.11
+opentelemetry-exporter-otlp-proto-http==1.26.0
+opentelemetry-exporter-otlp-proto-grpc==1.26.0
+kimina-client==0.2.1
+compressed-tensors==0.9.3
+opentelemetry-exporter-otlp==1.26.0
+vllm==0.8.5
+py-spy==0.4.1
+opencensus-context==0.1.3
+distlib==0.4.0
+colorful==0.5.8
+tensorboard-data-server==0.7.2
+google-auth==2.43.0
+google-api-core==2.28.1
+aiohttp-cors==0.8.1
+opentelemetry-exporter-prometheus==0.59b0
+opencensus==0.11.4
+verl==0.5.0.dev0
+opentelemetry-api==1.38.0
+opentelemetry-semantic-conventions==0.59b0
+opentelemetry-sdk==1.38.0
+typeguard==4.4.4
+shtab==1.8.0
+tyro==0.9.35
+verl==0.5.0.dev0
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2

wandb/run-20251218_182745-ahjkdup0/files/wandb-metadata.json

@@ -0,0 +1,92 @@
+{
+  "os": "Linux-5.14.0-284.25.1.el9_2.x86_64-x86_64-with-glibc2.35",
+  "python": "CPython 3.10.19",
+  "startedAt": "2025-12-18T10:27:45.627440Z",
+  "args": [
+    "--track"
+  ],
+  "program": "/mnt/general/wanghy/RAGEN/cleanrl/cleanrl/ppo_rubikscube.py",
+  "codePath": "cleanrl/cleanrl/ppo_rubikscube.py",
+  "codePathLocal": "cleanrl/cleanrl/ppo_rubikscube.py",
+  "git": {
+    "remote": "https://github.com/mll-lab-nu/RAGEN.git",
+    "commit": "8d73639b99b38265453f898b8d6af7d4af50d56e"
+  },
+  "email": "haoyu-wa22@mails.tsinghua.edu.cn",
+  "root": "/mnt/general/wanghy/RAGEN",
+  "host": "pt-4d654cf4576f4d23ad3d3919f12932fe-worker-0",
+  "executable": "/root/local/miniconda3/envs/ragen/bin/python",
+  "cpu_count": 64,
+  "cpu_count_logical": 128,
+  "gpu": "NVIDIA H100 80GB HBM3",
+  "gpu_count": 8,
+  "disk": {
+    "/": {
+      "total": "5153960755200",
+      "used": "447457550336"
+    }
+  },
+  "memory": {
+    "total": "2163642122240"
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-35e2d43d-4067-82ce-90d4-def9e389bf28"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-af4135e3-88f2-e9ac-518d-502c75a85429"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-d7fdeeba-fe9b-ec03-d9f7-6724fe4266b5"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-ccc4f668-3882-5a8e-2c07-c5cd08f6f666"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-7b73c0cf-d3d5-e10c-7176-a43be1e41001"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-81b58d94-5d1f-8ec2-f9d2-fd56172ed177"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-03e8bc66-3b44-6794-49fd-5392fbdda6d1"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-b7cf0ec6-7c29-1179-dceb-09565da51890"
+    }
+  ],
+  "cudaVersion": "12.4",
+  "writerId": "fcxjy4miz5x8bvnkek48fociofug1x4f"
+}

wandb/run-20251218_182745-ahjkdup0/files/wandb-summary.json

@@ -0,0 +1 @@
+{"eval/std_return":0.40424860853192807,"train/clipfrac":0,"charts/avg_value":0.4239591956138611,"losses/policy_loss":-8.033588528633118e-05,"train/success_rate":0.7542372881355932,"train/approx_kl":4.377216100692749e-08,"train/learning_rate":2.561475409836034e-07,"perf/SPS":2950,"train/policy_loss":-8.033588528633118e-05,"losses/explained_variance":0.6958094239234924,"rollout/ep_len_mean":11,"rollout/success_rate":1,"charts/train_success_rate":0.7542372881355932,"train/old_approx_kl":-4.877801984548569e-06,"losses/value_loss":0.03050171211361885,"global_step":999424,"train/value_loss":0.03050171211361885,"charts/progress":100,"train/entropy":0.6172537803649902,"_step":999424,"_runtime":348.911415774,"_wandb":{"runtime":348},"rollout/ep_rew_mean":1,"charts/SPS":2950,"_timestamp":1.7660540160092895e+09,"charts/avg_reward":0.0869140625,"eval/success_rate":0.79425,"eval/avg_return":0.79425,"eval/episodes":4000}

wandb/run-20251218_182745-ahjkdup0/run-ahjkdup0.wandb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:67d1c39f73bb838164f076d7389ae3af053f360a183c4ecef8eeaffc6cb81c2b
+size 35153374

wandb/run-20251218_183419-wvypkl8w/files/code/cleanrl/cleanrl/ppo_rubikscube.py

@@ -0,0 +1,517 @@
+# PPO with small MLP for RAGEN Rubik's Cube 2x2 using the existing env (no env edits)
+import os
+import random
+import time
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Tuple, Dict, Any, List
+import json
+import re
+
+import gymnasium as gym
+import numpy as np
+import torch
+import torch.nn as nn
+import torch.optim as optim
+import tyro
+from torch.distributions.categorical import Categorical
+
+import sys
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), '../../'))
+
+from ragen.env.rubikscube.env import RubiksCube2x2Env
+from ragen.env.rubikscube.config import RubiksCube2x2Config
+
+
+class RubiksCubeWrapper(gym.Env):
+    """
+    Adapter to use ragen RubiksCube2x2Env with Gymnasium vector API.
+    - Converts text observation to one-hot vector of 24 stickers x 6 colors.
+    - Maps agent actions [0..11] to env actions [1..12].
+    - Exposes proper observation_space and action_space.
+    """
+    metadata = {"render_modes": ["rgb_array", "human", "ansi"]}
+
+    def __init__(self, env: RubiksCube2x2Env):
+        super().__init__()
+        self._env = env
+        # 24 stickers, 6 colors -> one-hot size 144
+        self._colors = ['W', 'O', 'G', 'R', 'B', 'Y']
+        self._color_to_idx = {c: i for i, c in enumerate(self._colors)}
+        self.observation_space = gym.spaces.Box(low=0.0, high=1.0, shape=(24 * len(self._colors),), dtype=np.float32)
+        self.action_space = gym.spaces.Discrete(12)
+        # precompile regex to extract 4 letters within [X, X]\n [X, X]
+        # Lines look like: "Up (U):    [W, W]\n       [W, W]"
+        self._face_pat = re.compile(r"\[(?:\s*([A-Z])\s*,\s*([A-Z])\s*\])\n\s*\[(?:\s*([A-Z])\s*,\s*([A-Z])\s*\])")
+
+    def _encode_obs(self, text_obs: str) -> np.ndarray:
+        # Extract the six face blocks in the order U, L, F, R, B, D as rendered
+        faces_order = ["Up (U):", "Left (L):", "Front (F):", "Right (R):", "Back (B):", "Down (D):"]
+        onehots: List[int] = []
+        # Build a map from header to its block text
+        lines = text_obs.splitlines()
+        # Collect blocks starting after header line and including the next line for second row
+        i = 0
+        blocks: List[str] = []
+        while i < len(lines):
+            line = lines[i]
+            for header in faces_order:
+                if line.startswith(header):
+                    # Join current line's bracketed pair and the next line (which contains the second pair)
+                    # Remove the "Header:" prefix to keep only the bracket portions
+                    content = line[len(header):].strip()
+                    next_line = lines[i + 1] if i + 1 < len(lines) else ""
+                    block = f"{content}\n{next_line}"
+                    blocks.append(block)
+                    break
+            i += 1
+        # Fallback: if regex fails, return zeros
+        if len(blocks) != 6:
+            return np.zeros(24 * len(self._colors), dtype=np.float32)
+        stickers: List[int] = []
+        for blk in blocks:
+            m = self._face_pat.search(blk)
+            if not m:
+                return np.zeros(24 * len(self._colors), dtype=np.float32)
+            # order: 0,1,2,3 per render() doc
+            c0, c1, c2, c3 = m.group(1), m.group(2), m.group(3), m.group(4)
+            stickers.extend([c0, c1, c2, c3])
+        # stickers now length 24; convert to one-hot
+        grid = np.zeros((24, len(self._colors)), dtype=np.float32)
+        for idx, ch in enumerate(stickers):
+            cidx = self._color_to_idx.get(ch, None)
+            if cidx is not None:
+                grid[idx, cidx] = 1.0
+        return grid.reshape(-1)
+
+    def reset(self, *, seed: int | None = None, options: Dict[str, Any] | None = None):
+        text_obs = self._env.reset(seed=seed)
+        obs = self._encode_obs(text_obs)
+        return obs, {}
+
+    def step(self, action: int):
+        mapped = int(action) + 1  # 0..11 -> 1..12
+        text_obs, reward, done, info = self._env.step(mapped)
+        obs = self._encode_obs(text_obs)
+        terminated = bool(done)
+        truncated = False
+        return obs, float(reward), terminated, truncated, info or {}
+
+    def render(self):
+        return self._env.render()
+
+    def close(self):
+        self._env.close()
+
+
+@dataclass
+class Args:
+    exp_name: str = os.path.basename(__file__)[: -len(".py")]
+    seed: int = 1
+    torch_deterministic: bool = True
+    cuda: bool = True
+    track: bool = True
+    wandb_project_name: str = "cleanRL"
+    wandb_entity: str | None = None
+    capture_video: bool = False
+
+    # Algorithm
+    env_id: str = "RubiksCube2x2"
+    total_timesteps: int = 1000_000
+    learning_rate: float = 2.5e-4
+    num_envs: int = 8
+    num_steps: int = 128
+    anneal_lr: bool = True
+    gamma: float = 0.99
+    gae_lambda: float = 0.95
+    num_minibatches: int = 4
+    update_epochs: int = 4
+    norm_adv: bool = True
+    clip_coef: float = 0.2
+    clip_vloss: bool = True
+    ent_coef: float = 0.01
+    vf_coef: float = 0.5
+    max_grad_norm: float = 0.5
+    target_kl: float | None = None
+
+    # Rubik specific
+    scramble_depth: int = 4
+    max_steps_env: int = 20
+
+    # runtime filled
+    batch_size: int = 0
+    minibatch_size: int = 0
+    num_iterations: int = 0
+    eval_splits: int = 2
+    eval_episodes: int = 4000
+
+
+def make_env(idx, run_name, seed, scramble_depth, max_steps_env, capture_video=False):
+    def thunk():
+        config = RubiksCube2x2Config(scramble_depth=scramble_depth, max_steps=max_steps_env, render_mode='text')
+        env = RubiksCube2x2Env(config)
+        env = RubiksCubeWrapper(env)
+        env = gym.wrappers.TimeLimit(env, max_episode_steps=max_steps_env)
+        env = gym.wrappers.RecordEpisodeStatistics(env)
+        if capture_video and idx == 0:
+            env = gym.wrappers.RecordVideo(env, f"videos/{run_name}")
+        return env
+    return thunk
+
+
+def layer_init(layer, std=np.sqrt(2), bias_const=0.0):
+    torch.nn.init.orthogonal_(layer.weight, std)
+    torch.nn.init.constant_(layer.bias, bias_const)
+    return layer
+
+
+class Agent(nn.Module):
+    def __init__(self, envs):
+        super().__init__()
+        obs_shape = int(np.array(envs.single_observation_space.shape).prod())
+        hidden = 128
+        self.critic = nn.Sequential(
+            layer_init(nn.Linear(obs_shape, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, 1), std=1.0),
+        )
+        self.actor = nn.Sequential(
+            layer_init(nn.Linear(obs_shape, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, envs.single_action_space.n), std=0.01),
+        )
+
+    def get_value(self, x):
+        return self.critic(x)
+
+    def get_action_and_value(self, x, action=None):
+        logits = self.actor(x)
+        probs = Categorical(logits=logits)
+        if action is None:
+            action = probs.sample()
+        return action, probs.log_prob(action), probs.entropy(), self.critic(x)
+
+
+if __name__ == "__main__":
+    args = tyro.cli(Args)
+    args.batch_size = int(args.num_envs * args.num_steps)
+    args.minibatch_size = int(args.batch_size // args.num_minibatches)
+    args.num_iterations = args.total_timesteps // args.batch_size
+    run_name = f"{args.env_id}__{args.exp_name}__{args.seed}__{int(time.time())}"
+
+    if args.track:
+        import wandb
+        wandb.init(
+            project=args.wandb_project_name,
+            entity=args.wandb_entity,
+            config=vars(args),
+            name=run_name,
+            monitor_gym=True,
+            save_code=True,
+        )
+        try:
+            wandb.define_metric("global_step")
+            for prefix in ["train/*", "rollout/*", "eval/*", "losses/*", "charts/*", "perf/*"]:
+                wandb.define_metric(prefix, step_metric="global_step")
+        except Exception:
+            pass
+
+    # seeding
+    random.seed(args.seed)
+    np.random.seed(args.seed)
+    torch.manual_seed(args.seed)
+    torch.backends.cudnn.deterministic = args.torch_deterministic
+
+    device = torch.device("cuda" if torch.cuda.is_available() and args.cuda else "cpu")
+
+    # envs
+    envs = gym.vector.SyncVectorEnv([
+        make_env(i, run_name, args.seed, args.scramble_depth, args.max_steps_env, args.capture_video)
+        for i in range(args.num_envs)
+    ])
+    assert isinstance(envs.single_action_space, gym.spaces.Discrete)
+
+    agent = Agent(envs).to(device)
+    optimizer = optim.Adam(agent.parameters(), lr=args.learning_rate, eps=1e-5)
+
+    # storage
+    obs = torch.zeros((args.num_steps, args.num_envs) + envs.single_observation_space.shape).to(device)
+    actions = torch.zeros((args.num_steps, args.num_envs) + envs.single_action_space.shape).to(device)
+    logprobs = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    rewards = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    dones = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    values = torch.zeros((args.num_steps, args.num_envs)).to(device)
+
+    # start
+    global_step = 0
+    start_time = time.time()
+    next_obs, _ = envs.reset(seed=args.seed)
+    next_obs = torch.Tensor(next_obs).to(device)
+    next_done = torch.zeros(args.num_envs).to(device)
+
+    # eval helper similar to FrozenLake: collect greedy eval trajectories and write metrics.json
+    def collect_eval_trajectories(agent_model, make_env_fn, n_episodes, step_tag):
+        out_dir = Path(f"runs/{run_name}/trajectories/step_{step_tag}")
+        out_dir.mkdir(parents=True, exist_ok=True)
+        out_path = out_dir / "trajectories.jsonl"
+        env = make_env_fn()
+        collected = 0
+        summary_returns = []
+        summary_success = []
+        with out_path.open("w") as f:
+            while collected < n_episodes:
+                state, _ = env.reset(seed=args.seed + 100000 + collected)
+                traj_states = [state.tolist()]
+                traj_actions = []
+                traj_rewards = []
+                traj_dones = []
+                traj_success = []
+                done = False
+                step_count = 0
+                max_eval_steps = getattr(env, '_max_episode_steps', None) or int(args.max_steps_env)
+                while not done:
+                    with torch.no_grad():
+                        logits = agent_model.actor(torch.tensor(state, dtype=torch.float32, device=device).unsqueeze(0))
+                        action = int(torch.argmax(logits, dim=1).item())
+                    next_state, reward, terminated, truncated, info = env.step(action)
+                    traj_actions.append(int(action))
+                    traj_rewards.append(float(reward))
+                    step_count += 1
+                    d = bool(terminated) or bool(truncated) or (step_count >= max_eval_steps)
+                    traj_dones.append(d)
+                    traj_success.append(bool((info or {}).get('success', False)))
+                    state = next_state
+                    traj_states.append(state.tolist())
+                    done = d
+                ep_ret = float(sum(traj_rewards))
+                ep_succ = bool(any(traj_success))
+                record = {
+                    "states": traj_states,
+                    "actions": traj_actions,
+                    "rewards": traj_rewards,
+                    "dones": traj_dones,
+                    "success": traj_success,
+                    "episode_return": ep_ret,
+                    "episode_success": ep_succ,
+                }
+                f.write(json.dumps(record) + "\n")
+                collected += 1
+                summary_returns.append(ep_ret)
+                summary_success.append(1.0 if ep_succ else 0.0)
+        env.close()
+        try:
+            metrics = {
+                "global_step": int(step_tag),
+                "episodes": int(n_episodes),
+                "success_rate": float(np.mean(summary_success)) if len(summary_success) else 0.0,
+                "avg_return": float(np.mean(summary_returns)) if len(summary_returns) else 0.0,
+                "std_return": float(np.std(summary_returns)) if len(summary_returns) else 0.0,
+            }
+            with (out_dir / "metrics.json").open("w") as mf:
+                json.dump(metrics, mf)
+        except Exception as e:
+            print(f"Warning: failed to write eval metrics: {e}")
+
+    # training loop
+    eval_every_iters = max(1, args.num_iterations // args.eval_splits)
+    for iteration in range(1, args.num_iterations + 1):
+        # Anneal LR
+        if args.anneal_lr:
+            frac = 1.0 - (iteration - 1.0) / args.num_iterations
+            lrnow = frac * args.learning_rate
+            optimizer.param_groups[0]["lr"] = lrnow
+
+        # accumulate per-iteration episode successes
+        iter_successes = []
+        for step in range(0, args.num_steps):
+            global_step += args.num_envs
+            obs[step] = next_obs
+            dones[step] = next_done
+
+            with torch.no_grad():
+                action, logprob, _, value = agent.get_action_and_value(next_obs)
+                values[step] = value.flatten()
+            actions[step] = action
+            logprobs[step] = logprob
+
+            next_obs, reward, terminations, truncations, infos = envs.step(action.cpu().numpy())
+            next_done = np.logical_or(terminations, truncations)
+            rewards[step] = torch.tensor(reward).to(device).view(-1)
+            next_obs, next_done = torch.Tensor(next_obs).to(device), torch.Tensor(next_done).to(device)
+
+            # Episode stats logging similar to FrozenLake
+            try:
+                mask = None
+                if isinstance(infos, dict):
+                    if "_episode" in infos:
+                        mask = np.asarray(infos["_episode"]).astype(bool)
+                    elif "episode" in infos and isinstance(infos["episode"], dict) and "_l" in infos["episode"]:
+                        mask = np.asarray(infos["episode"]["_l"]).astype(bool)
+                if mask is not None and np.any(mask):
+                    r_arr = np.asarray(infos.get("episode", {}).get("r", np.zeros_like(mask, dtype=float)))
+                    l_arr = np.asarray(infos.get("episode", {}).get("l", np.zeros_like(mask, dtype=int)))
+                    # prefer success from info; fallback to ep return > 0
+                    if "success" in infos:
+                        succ_arr = np.asarray(infos.get("success", np.zeros_like(mask, dtype=bool))).astype(float)
+                    else:
+                        try:
+                            succ_arr = (np.asarray(r_arr) > 0).astype(float)
+                        except Exception:
+                            succ_arr = np.zeros_like(mask, dtype=float)
+                    # collect iteration successes for training success rate
+                    try:
+                        for s in np.asarray(succ_arr)[mask]:
+                            iter_successes.append(float(s))
+                    except Exception:
+                        pass
+                    if args.track:
+                        try:
+                            import wandb
+                            log_dict = {
+                                "global_step": int(global_step),
+                                "rollout/ep_rew_mean": float(np.mean(r_arr[mask])) if np.any(mask) else None,
+                                "rollout/ep_len_mean": float(np.mean(l_arr[mask])) if np.any(mask) else None,
+                                "rollout/success_rate": float(np.mean(succ_arr[mask])) if np.any(mask) else None,
+                            }
+                            wandb.log(log_dict, step=global_step)
+                        except Exception:
+                            pass
+            except Exception:
+                pass
+
+        # GAE
+        with torch.no_grad():
+            next_value = agent.get_value(next_obs).reshape(1, -1)
+            advantages = torch.zeros_like(rewards).to(device)
+            lastgaelam = 0
+            for t in reversed(range(args.num_steps)):
+                if t == args.num_steps - 1:
+                    nextnonterminal = 1.0 - next_done
+                    nextvalues = next_value
+                else:
+                    nextnonterminal = 1.0 - dones[t + 1]
+                    nextvalues = values[t + 1]
+                delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
+                advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
+            returns = advantages + values
+
+        # flatten batch
+        b_obs = obs.reshape((-1,) + envs.single_observation_space.shape)
+        b_logprobs = logprobs.reshape(-1)
+        b_actions = actions.reshape((-1,) + envs.single_action_space.shape)
+        b_advantages = advantages.reshape(-1)
+        b_returns = returns.reshape(-1)
+        b_values = values.reshape(-1)
+
+        # update
+        b_inds = np.arange(args.batch_size)
+        clipfracs = []
+        for epoch in range(args.update_epochs):
+            np.random.shuffle(b_inds)
+            for start in range(0, args.batch_size, args.minibatch_size):
+                end = start + args.minibatch_size
+                mb_inds = b_inds[start:end]
+
+                _, newlogprob, entropy, newvalue = agent.get_action_and_value(b_obs[mb_inds], b_actions.long()[mb_inds])
+                logratio = newlogprob - b_logprobs[mb_inds]
+                ratio = logratio.exp()
+
+                with torch.no_grad():
+                    old_approx_kl = (-logratio).mean()
+                    approx_kl = ((ratio - 1) - logratio).mean()
+                    clipfracs += [((ratio - 1.0).abs() > args.clip_coef).float().mean().item()]
+
+                mb_advantages = b_advantages[mb_inds]
+                if args.norm_adv:
+                    mb_advantages = (mb_advantages - mb_advantages.mean()) / (mb_advantages.std() + 1e-8)
+
+                pg_loss1 = -mb_advantages * ratio
+                pg_loss2 = -mb_advantages * torch.clamp(ratio, 1 - args.clip_coef, 1 + args.clip_coef)
+                pg_loss = torch.max(pg_loss1, pg_loss2).mean()
+
+                newvalue = newvalue.view(-1)
+                if args.clip_vloss:
+                    v_loss_unclipped = (newvalue - b_returns[mb_inds]) ** 2
+                    v_clipped = b_values[mb_inds] + torch.clamp(
+                        newvalue - b_values[mb_inds], -args.clip_coef, args.clip_coef,
+                    )
+                    v_loss_clipped = (v_clipped - b_returns[mb_inds]) ** 2
+                    v_loss = 0.5 * torch.max(v_loss_unclipped, v_loss_clipped).mean()
+                else:
+                    v_loss = 0.5 * ((newvalue - b_returns[mb_inds]) ** 2).mean()
+
+                entropy_loss = entropy.mean()
+                loss = pg_loss - args.ent_coef * entropy_loss + v_loss * args.vf_coef
+
+                optimizer.zero_grad()
+                loss.backward()
+                nn.utils.clip_grad_norm_(agent.parameters(), args.max_grad_norm)
+                optimizer.step()
+
+            if args.target_kl is not None and approx_kl > args.target_kl:
+                break
+
+        # metrics
+        y_pred, y_true = b_values.cpu().numpy(), b_returns.cpu().numpy()
+        var_y = np.var(y_true)
+        explained_var = np.nan if var_y == 0 else 1 - np.var(y_true - y_pred) / var_y
+
+        sps = int(global_step / (time.time() - start_time))
+        train_success_rate = float(np.mean(iter_successes)) if len(iter_successes) else 0.0
+        print(f"Iter {iteration:4d}/{args.num_iterations} | SPS: {sps:5d} | R: {rewards.mean().item():6.3f}")
+        if args.track:
+            try:
+                import wandb
+                wandb.log({
+                    "global_step": int(global_step),
+                    "charts/progress": float(100.0 * iteration / max(1, args.num_iterations)),
+                    "train/value_loss": float(v_loss.item()),
+                    "train/policy_loss": float(pg_loss.item()),
+                    "losses/value_loss": float(v_loss.item()),
+                    "losses/policy_loss": float(pg_loss.item()),
+                    "train/entropy": float(entropy_loss.item()),
+                    "train/old_approx_kl": float(old_approx_kl.item()),
+                    "train/approx_kl": float(approx_kl.item()),
+                    "train/clipfrac": float(np.mean(clipfracs)) if len(clipfracs) else 0.0,
+                    "losses/explained_variance": float(explained_var),
+                    "charts/avg_reward": float(rewards.mean().item()),
+                    "charts/avg_value": float(values.mean().item()),
+                    "perf/SPS": int(sps),
+                    "charts/SPS": int(sps),
+                    "train/success_rate": train_success_rate,
+                    "charts/train_success_rate": train_success_rate,
+                    "train/learning_rate": float(optimizer.param_groups[0]["lr"]),
+                }, step=global_step)
+            except Exception:
+                pass
+
+        # periodic evaluation collection
+        if iteration % eval_every_iters == 0:
+            try:
+                eval_thunk = make_env(0, run_name, args.seed + 9999, args.scramble_depth, args.max_steps_env, False)
+                collect_eval_trajectories(agent, eval_thunk, n_episodes=args.eval_episodes, step_tag=global_step)
+                if args.track:
+                    try:
+                        import json as _json
+                        from pathlib import Path as _Path
+                        mpath = _Path(f"runs/{run_name}/trajectories/step_{global_step}/metrics.json")
+                        if mpath.exists():
+                            with mpath.open("r") as mf:
+                                metrics = _json.load(mf)
+                            wandb.log({
+                                "eval/success_rate": metrics.get("success_rate"),
+                                "eval/avg_return": metrics.get("avg_return"),
+                                "eval/std_return": metrics.get("std_return"),
+                                "eval/episodes": metrics.get("episodes"),
+                            }, step=global_step)
+                    except Exception:
+                        pass
+                print(f"Collected {args.eval_episodes} eval trajectories at global_step {global_step}")
+            except Exception as e:
+                print(f"Warning: eval trajectory collection failed at step {global_step}: {e}")
+
+    envs.close()

wandb/run-20251218_183419-wvypkl8w/files/config.yaml

@@ -0,0 +1,201 @@
+_wandb:
+    value:
+        cli_version: 0.23.0
+        code_path: code/cleanrl/cleanrl/ppo_rubikscube.py
+        e:
+            bay0zucrvgh0budvh5sfo6odu02gecuh:
+                args:
+                    - --track
+                codePath: cleanrl/cleanrl/ppo_rubikscube.py
+                codePathLocal: cleanrl/cleanrl/ppo_rubikscube.py
+                cpu_count: 64
+                cpu_count_logical: 128
+                cudaVersion: "12.4"
+                disk:
+                    /:
+                        total: "5153960755200"
+                        used: "447541800960"
+                email: haoyu-wa22@mails.tsinghua.edu.cn
+                executable: /root/local/miniconda3/envs/ragen/bin/python
+                git:
+                    commit: 8d73639b99b38265453f898b8d6af7d4af50d56e
+                    remote: https://github.com/mll-lab-nu/RAGEN.git
+                gpu: NVIDIA H100 80GB HBM3
+                gpu_count: 8
+                gpu_nvidia:
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-35e2d43d-4067-82ce-90d4-def9e389bf28
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-af4135e3-88f2-e9ac-518d-502c75a85429
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-d7fdeeba-fe9b-ec03-d9f7-6724fe4266b5
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-ccc4f668-3882-5a8e-2c07-c5cd08f6f666
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-7b73c0cf-d3d5-e10c-7176-a43be1e41001
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-81b58d94-5d1f-8ec2-f9d2-fd56172ed177
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-03e8bc66-3b44-6794-49fd-5392fbdda6d1
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-b7cf0ec6-7c29-1179-dceb-09565da51890
+                host: pt-4d654cf4576f4d23ad3d3919f12932fe-worker-0
+                memory:
+                    total: "2163642122240"
+                os: Linux-5.14.0-284.25.1.el9_2.x86_64-x86_64-with-glibc2.35
+                program: /mnt/general/wanghy/RAGEN/cleanrl/cleanrl/ppo_rubikscube.py
+                python: CPython 3.10.19
+                root: /mnt/general/wanghy/RAGEN
+                startedAt: "2025-12-18T10:34:19.463187Z"
+                writerId: bay0zucrvgh0budvh5sfo6odu02gecuh
+        m:
+            - "1": global_step
+              "6":
+                - 3
+              "7": []
+            - "2": losses/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": charts/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": perf/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": train/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": rollout/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": eval/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+        python_version: 3.10.19
+        t:
+            "1":
+                - 1
+                - 11
+                - 30
+                - 49
+                - 50
+                - 51
+                - 105
+            "2":
+                - 1
+                - 11
+                - 30
+                - 49
+                - 50
+                - 51
+                - 105
+            "3":
+                - 7
+                - 13
+                - 16
+                - 61
+            "4": 3.10.19
+            "5": 0.23.0
+            "6": 4.57.1
+            "12": 0.23.0
+            "13": linux-x86_64
+anneal_lr:
+    value: true
+batch_size:
+    value: 1024
+capture_video:
+    value: false
+clip_coef:
+    value: 0.2
+clip_vloss:
+    value: true
+cuda:
+    value: true
+ent_coef:
+    value: 0.01
+env_id:
+    value: RubiksCube2x2
+eval_episodes:
+    value: 4000
+eval_splits:
+    value: 2
+exp_name:
+    value: ppo_rubikscube
+gae_lambda:
+    value: 0.95
+gamma:
+    value: 0.99
+learning_rate:
+    value: 0.00025
+max_grad_norm:
+    value: 0.5
+max_steps_env:
+    value: 20
+minibatch_size:
+    value: 256
+norm_adv:
+    value: true
+num_envs:
+    value: 8
+num_iterations:
+    value: 976
+num_minibatches:
+    value: 4
+num_steps:
+    value: 128
+scramble_depth:
+    value: 4
+seed:
+    value: 1
+target_kl:
+    value: null
+torch_deterministic:
+    value: true
+total_timesteps:
+    value: 1000000
+track:
+    value: true
+update_epochs:
+    value: 4
+vf_coef:
+    value: 0.5
+wandb_entity:
+    value: null
+wandb_project_name:
+    value: cleanRL

wandb/run-20251218_183419-wvypkl8w/files/diff.patch

@@ -0,0 +1,1162 @@
+diff --git a/config/base.yaml b/config/base.yaml
+index a1659e4..9df6250 100644
+--- a/config/base.yaml
++++ b/config/base.yaml
+@@ -9,15 +9,16 @@ seed:
+   train: 10000
+   val: 123
+ 
+-micro_batch_size_per_gpu: 4
++micro_batch_size_per_gpu: 1
+ ppo_mini_batch_size: 32
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/Qwen/Qwen2.5-0.5B-Instruct
++# Qwen/Qwen2.5-0.5B-Instruct
+ enable_response_mask: True # Enabling response mask could improve stability of rollout/old_log_prob, as P(st|history) are no longer calculated in loss here. See https://docs.google.com/document/d/1bg7obeiKTExuHHBl5uOiSpec5uLDZ2Tgvxy6li5pHX4/edit?usp=sharing for more details.
+ grpo_advantage_length_weight: False # if you do not enable this and critic/advantage_estimator is GRPO, and the critic/advantages/mean is too low, then you can try enabling this to encourage reasoning and forbid collapse
+ 
+ lora:
+   rank: 0
+-  alpha: 64
++  alpha: 16
+   target_modules: all-linear
+ 
+ actor_rollout_ref:
+@@ -46,10 +47,10 @@ actor_rollout_ref:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu} # following micro_batch_size_per_gpu
+     tensor_model_parallel_size: 1
+-    max_model_len: 3600
++    max_model_len: 7200 #3600 why** 14400
+     prompt_length: 1 # useless. Just put it here
+     response_length: 400 # single-turn response length
+-    gpu_memory_utilization: 0.5
++    gpu_memory_utilization: 0.7
+     max_num_batched_tokens: 8192 # set only when enable_chunked_prefill is true
+     temperature: 1
+     rollout_filter_ratio: 0.25
+@@ -90,26 +91,28 @@ algorithm:
+     kl_coef: 0.000
+ 
+ trainer:
+-  project_name: ragen_latest
++  project_name: 
+   experiment_name: test
+   local_log_dir: "results/"
+-  save_freq: 100
++  save_freq: -1
+   total_training_steps: 200
+   validation_steps: 1 # validation instances = validation_steps * val_env_groups * group_size
+   val_before_train: True
+   n_gpus_per_node: 1
+   test_freq: 10
+   generations_to_log_to_wandb: 
++    train: 128
+     val: 20
+   logger: [ 'console', 'wandb' ]
+   max_actor_ckpt_to_keep: 1
+   max_critic_ckpt_to_keep: 1
++  default_local_dir: /mnt/general/wanghy/RAGEN/saves/
+ 
+ agent_proxy:
+   max_context_window: -1 # set a value > 0 to enable context window for long trajectory
+-  max_turn: 5
++  max_turn: 25 #25 why** 700
+   action_sep: "||"
+-  max_actions_per_turn: 2 # how many actions can be output at most in a single turn
++  max_actions_per_turn: 1 # how many actions can be output at most in a single turn
+   use_turn_scores: False # important to GAE when applying token-level rewards to token-level advantages. If False, will take the sum of scores as the reward for the last turn.
+   enable_think: True # False -> no think RL
+   reward_normalization:
+@@ -126,11 +129,11 @@ es_manager:
+       tags: ["CoordSokoban"]
+       n_groups: [8] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+   val:
+-    env_groups: 32
+-    group_size: 16 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
++    env_groups: 256
++    group_size: 1 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
+     env_configs:
+       tags: ["CoordSokoban"]
+-      n_groups: [32] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      n_groups: [256] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ ctx_manager:
+   generation: # go to vllm
+diff --git a/config/envs.yaml b/config/envs.yaml
+index 1002ac8..ba3ac6f 100644
+--- a/config/envs.yaml
++++ b/config/envs.yaml
+@@ -2,6 +2,7 @@ custom_envs:
+   SimpleSokoban:
+     env_type: sokoban
+     max_actions_per_traj: 10 # used in environment state manager to control the actual max actions executed per trajectory
++    # ORIGNAL env_instruction
+     env_instruction: |
+       You are solving the Sokoban puzzle. 
+       You are the player and you need to push all boxes to targets. 
+@@ -12,7 +13,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+ 
+   LargerSokoban:
+@@ -65,7 +66,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+       observation_format: "grid_coord"
+ 
+@@ -111,14 +112,14 @@ custom_envs:
+ 
+   FrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: "You are solving the FrozenLake puzzle. Forbid the whole and go to the target. You may move to the unintended direction due to the slippery ice. Example answer format: <think>To forbid the hole and go to the target, I should go left then go up.</think><answer>Left || Up</answer>"
+     max_tokens: 100
+     env_config: null
+ 
+   CoordFrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: |
+       You are solving the FrozenLake puzzle. The observation includes both a symbol grid and zero-indexed coordinates for the start, goal, player, and any holes.
+       Coordinates range from the top-left corner (0, 0) to the bottom-right corner (5, 5).
+@@ -195,4 +196,80 @@ custom_envs:
+     max_actions_per_traj: 30
+     env_instruction: "You are a Lean theorem prover. Given a Lean theorem statement, propose a sequence of tactics that completes the proof. Think step by step about which tactics to apply next. Provide tactics separated by '||', for example <answer>intro || simp || rfl</answer>."
+     max_tokens: 512
+-    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
+
++    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
++
++  
++  game_2048:
++    env_type: game_2048
++    max_actions_per_traj: 700
++    env_instruction: |
++      You are playing the 2048 game on a 4x4 grid. Merge equal tiles by sliding Up, Right, Down, or Left.
++      If a move is invalid (no tiles move), a small penalty is applied. Respond with a single action.
++      Example: <answer>Up</answer>
++    max_tokens: 8192
++    env_config: null
++
++  blackjack:
++    env_type: blackjack
++    max_actions_per_traj: 10
++    env_instruction: |
++      You are playing Blackjack against a dealer. The dealer must hit on 16 or less and stand on 17 or more.
++      Choose either Stick or Hit. Respond with a single action.
++      Example: <answer>Hit</answer>
++    max_tokens: 64
++    env_config: null
++
++  rubikscube:
++    env_type: rubikscube
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.
++      Available actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.
++      - Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).
++      - Modifiers: A letter alone means 90° clockwise (e.g., 'R'). A letter with prime (') means 90° counter-clockwise (e.g., "R'").
++      Respond with a sequence of actions separated by "||".
++      Example: <answer>U</answer>
++    max_tokens: 96
++    env_config:
++      scramble_depth: 1
++      max_steps: 20
++      render_mode: "text"
++  
++  SimpleSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 2x2 box contains the numbers 1-4 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format, for example: <answer>place 1 at row 2 col 3</answer> or <answer>1,2,3</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 4
++      difficulty: "easy"
++      render_format: "with_feedback"
++      show_conflicts: false
++      show_valid_numbers: false
++      max_steps: 20
++
++  MediumSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 30
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 3x3 box contains the numbers 1-9 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format: <answer>place 5 at row 2 col 3</answer> or <answer>2,3,5</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 9
++      difficulty: "medium"
++      render_format: "with_feedback"
++      show_conflicts: true
++      show_valid_numbers: true
++      max_steps: 81
++
+diff --git a/config/eval.yaml b/config/eval.yaml
+index e65f3a5..7775fd1 100644
+--- a/config/eval.yaml
++++ b/config/eval.yaml
+@@ -1,14 +1,15 @@
+ defaults:
+   - envs
+-
++enable_response_mask: True
+ system:
+-  CUDA_VISIBLE_DEVICES: "0"
++  CUDA_VISIBLE_DEVICES: "0,1,2,3,4,5,6,7"
+ 
+ seed:
+   train: 10000
+   val: 123
+ 
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/openai/gpt-oss-20b
++# /mnt/general/share/model/tyzhu/SPA-frozenlake-qwen2.5-1.5b-instruct
+ 
+ lora:
+   rank: 0
+@@ -24,9 +25,9 @@ actor_rollout_ref:
+   rollout:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu}
+-    tensor_model_parallel_size: 1
++    tensor_model_parallel_size: 8
+     dtype: bfloat16
+-    max_model_len: 3600
++    max_model_len: 7200
+     prompt_length: 1
+     response_length: 400
+     gpu_memory_utilization: 0.9
+@@ -35,6 +36,7 @@ actor_rollout_ref:
+     free_cache_engine: True
+     enable_chunked_prefill: False
+     disable_log_stats: False
++    do_sample: True
+     val_kwargs:
+       do_sample: True
+       temperature: 0.5
+@@ -44,9 +46,9 @@ actor_rollout_ref:
+ 
+ agent_proxy:
+   max_context_window: -1
+-  max_turn: 5
++  max_turn: 25
+   action_sep: "||"
+-  max_actions_per_turn: 2
++  max_actions_per_turn: 1
+   use_turn_scores: False
+   enable_think: True
+   reward_normalization:
+@@ -59,13 +61,13 @@ es_manager:
+     env_groups: 8
+     group_size: 16
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [8]
+   val:
+     env_groups: 32
+-    group_size: 16
++    group_size: 128
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [32]
+ 
+ ctx_manager:
+diff --git a/config/evaluate_api_llm.yaml b/config/evaluate_api_llm.yaml
+index b21c95e..e2e6761 100644
+--- a/config/evaluate_api_llm.yaml
++++ b/config/evaluate_api_llm.yaml
+@@ -1,8 +1,11 @@
++#export OPENAI_BASE_URL="https://api.ohmygpt.com/v1"
++#export OPENAI_API_KEY="sk-o4sMxBkN5BB100C4D4a3T3BlBkFJF7791CA39EA14ca98041"
++#python -m ragen.eval_api hydra.searchpath='[file://./verl/verl/trainer/config]'
+ defaults:
+   - base # this is a symbolic link to the verl/verl/trainer/config/ppo_trainer.yaml file
+ 
+ model_config:
+-  model_name: gpt-4o # should be registered in model_info
++  model_name: ark-deepseek-v3-250324 # should be registered in model_info
+   max_concurrency: 16
+ 
+ model_info:
+@@ -24,9 +27,9 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+-  gpt-4o:
++  gpt-4o-mini:
+     provider_name: openai
+-    model_name: gpt-4o
++    model_name: gpt-4o-mini
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+@@ -36,21 +39,42 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  ark-deepseek-v3-250324:
++    provider_name: openai
++    model_name: ark-deepseek-v3-250324
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
+   deepseek-v3:
+     provider_name: deepseek
+     model_name: deepseek-chat
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  glm-4.6:
++    provider_name: openai
++    model_name: glm-4.6
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
++  TA/openai/gpt-oss-120b:
++    provider_name: openai
++    model_name: TA/openai/gpt-oss-120b
++    generation_kwargs:
++      temperature: 0
++      max_tokens: 8192
++      # max_retries: 5
+ 
+-
+-
++agent_proxy:
++  max_turn: 5
+ es_manager:
+   val:
+-    env_groups: 256
++    env_groups: 128
+     group_size: 1 # should be set to 1 because val temperature is set to 0 and same prompt leads to same output
+     env_configs:
+-      tags: ["CoordSokoban"]
+-      n_groups: [256] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      tags: ["rubikscube"]
++      n_groups: [128] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ 
++rollout:
++  max_model_len: 7200
+
+Submodule external/webshop-minimal contains modified content
+diff --git a/external/webshop-minimal/requirements.txt b/external/webshop-minimal/requirements.txt
+index 5a1b04f..238ed5a 100644
+--- a/external/webshop-minimal/requirements.txt
++++ b/external/webshop-minimal/requirements.txt
+@@ -4,7 +4,7 @@ flask
+ html2text
+ rank_bm25
+ pyserini
+-faiss-cpu
++faiss-gpu
+ thefuzz
+ gdown
+ spacy
+diff --git a/ragen/env/__init__.py b/ragen/env/__init__.py
+index b0f3461..bc6838e 100644
+--- a/ragen/env/__init__.py
++++ b/ragen/env/__init__.py
+@@ -12,6 +12,14 @@ from .metamathqa.env import MetaMathQAEnv
+ from .metamathqa.config import MetaMathQAEnvConfig
+ from .lean.config import LeanEnvConfig
+ from .lean.env import LeanEnv
++from .game_2048.config import Game2048EnvConfig
++from .game_2048.env import Game2048Env
++from .blackjack.config import BlackjackEnvConfig
++from .blackjack.env import BlackjackEnv
++from .rubikscube.config import RubiksCube2x2Config
++from .rubikscube.env import RubiksCube2x2Env
++from .sudoku.config import SudokuEnvConfig
++from .sudoku.env import SudokuEnv
+ 
+ 
+ REGISTERED_ENVS = {
+@@ -22,6 +30,10 @@ REGISTERED_ENVS = {
+     # 'alfworld': AlfredTXTEnv,
+     'metamathqa': MetaMathQAEnv,
+     'lean': LeanEnv,
++    'game_2048': Game2048Env,
++    'blackjack': BlackjackEnv,
++    'rubikscube': RubiksCube2x2Env,
++    'sudoku': SudokuEnv,
+ }
+ 
+ REGISTERED_ENV_CONFIGS = {
+@@ -32,6 +44,10 @@ REGISTERED_ENV_CONFIGS = {
+     # 'alfworld': AlfredEnvConfig,
+     'metamathqa': MetaMathQAEnvConfig,
+     'lean': LeanEnvConfig,
++    'game_2048': Game2048EnvConfig,
++    'blackjack': BlackjackEnvConfig,
++    'rubikscube': RubiksCube2x2Config,
++    'sudoku': SudokuEnvConfig,
+ }
+ 
+ try:
+diff --git a/ragen/env/frozen_lake/config.py b/ragen/env/frozen_lake/config.py
+index de054f4..9950c34 100644
+--- a/ragen/env/frozen_lake/config.py
++++ b/ragen/env/frozen_lake/config.py
+@@ -8,7 +8,7 @@ class FrozenLakeEnvConfig:
+     size: int = 4
+     p: float = 0.9
+     success_rate: float = 0.8
+-    is_slippery: bool = True
++    is_slippery: bool = False
+     map_seed: Optional[int] = None
+     render_mode: str = "text"
+     observation_format: str = "grid"
+diff --git a/ragen/env/frozen_lake/env.py b/ragen/env/frozen_lake/env.py
+index 9ee3add..eef11d8 100644
+--- a/ragen/env/frozen_lake/env.py
++++ b/ragen/env/frozen_lake/env.py
+@@ -13,6 +13,7 @@ from ragen.env.base import BaseDiscreteActionEnv
+ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+     def __init__(self, config: FrozenLakeEnvConfig = FrozenLakeEnvConfig()):
+         # Using mappings directly from config
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.GRID_LOOKUP = config.grid_lookup
+         self.ACTION_LOOKUP = config.action_lookup
+@@ -95,6 +96,7 @@ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+ if __name__ == "__main__":
+     import matplotlib.pyplot as plt
+     config = FrozenLakeEnvConfig(size=4, is_slippery=True)
++    # import pdb;pdb.set_trace()
+     env = FrozenLakeEnv(config)
+     print(env.reset())
+     while True:
+diff --git a/ragen/env/sokoban/env.py b/ragen/env/sokoban/env.py
+index 17cd636..6922b54 100644
+--- a/ragen/env/sokoban/env.py
++++ b/ragen/env/sokoban/env.py
+@@ -33,6 +33,7 @@ class SokobanEnv(BaseDiscreteActionEnv, GymSokobanEnv):
+     def reset(self, seed=None, mode=None):
+         try:
+             with all_seed(seed):
++                # import pdb;pdb.set_trace()
+                 self.room_fixed, self.room_state, self.box_mapping, action_sequence = generate_room(
+                     dim=self.dim_room,
+                     num_steps=self.num_gen_steps,
+diff --git a/ragen/llm_agent/agent_proxy.py b/ragen/llm_agent/agent_proxy.py
+index c25b50d..f6c4088 100644
+--- a/ragen/llm_agent/agent_proxy.py
++++ b/ragen/llm_agent/agent_proxy.py
+@@ -287,6 +287,7 @@ def main(config):
+ 	proxy = LLMAgentProxy(config, actor_wg, tokenizer)
+ 	import time
+ 	start_time = time.time()
++	# import pdb;pdb.set_trace()
+ 	rollouts = proxy.rollout(
+ 		DataProto(
+ 			batch=None,
+diff --git a/ragen/llm_agent/base_llm.py b/ragen/llm_agent/base_llm.py
+index 358a2eb..b6f7797 100644
+--- a/ragen/llm_agent/base_llm.py
++++ b/ragen/llm_agent/base_llm.py
+@@ -33,6 +33,7 @@ class OpenAIProvider(LLMProvider):
+             raise ValueError("OpenAI API key not provided and not found in environment variables")
+         
+         self.client = AsyncOpenAI(api_key=self.api_key)
++        # import pdb;pdb.set_trace()
+     
+     async def generate(self, messages: List[Dict[str, str]], **kwargs) -> LLMResponse:
+         if "o1-mini" in self.model_name:
+@@ -46,10 +47,7 @@ class OpenAIProvider(LLMProvider):
+         )
+         if response.choices[0].finish_reason in ['length', 'content_filter']:
+             raise ValueError("Content filtered or length exceeded")
+-        return LLMResponse(
+-            content=response.choices[0].message.content,
+-            model_name=response.model
+-        )
++        return LLMResponse(content=response.choices[0].message.content,model_name=response.model)
+ 
+ class DeepSeekProvider(LLMProvider):
+     """DeepSeek API provider implementation"""
+@@ -199,7 +197,7 @@ class ConcurrentLLM:
+         
+         # Queue to store unfinished or failed tasks
+         current_batch = messages_list.copy()
+-        max_retries = kwargs.get("max_retries", 100)
++        max_retries = kwargs.get("max_retries", 10)
+         retry_count = 0
+         
+         while current_batch and retry_count < max_retries:
+diff --git a/ragen/llm_agent/ctx_manager.py b/ragen/llm_agent/ctx_manager.py
+index 905247a..af20a01 100644
+--- a/ragen/llm_agent/ctx_manager.py
++++ b/ragen/llm_agent/ctx_manager.py
+@@ -88,6 +88,7 @@ class ContextManager:
+         Initialize the ContextManager.
+         Processor is used to process the image data.
+         """
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.tokenizer = tokenizer
+         self.processor = processor
+@@ -311,7 +312,7 @@ class ContextManager:
+ 
+             llm_input_texts.append(text_with_prompt)
+             messages_list.append(messages)
+-
++        # import pdb;pdb.set_trace()
+         inputs = self.tokenizer(llm_input_texts, return_tensors="pt", padding=True, padding_side="left", truncation=False) # We have truncated previously, truncation in tokenizer may cause issues.
+         input_ids, attention_mask = inputs.input_ids, inputs.attention_mask
+         position_ids = (attention_mask.cumsum(dim=-1) - 1).clamp(min=0)
+@@ -353,6 +354,30 @@ class ContextManager:
+                 key: np.sum(value) / self.env_nums[key.split("/")[0]]
+                 for key, value in metrics.items()
+             }
++            # Derived metrics for wandb logging
++            try:
++                # charts/avg_episode_return: average across all 2048 env groups
++                ep_keys = [k for k in metrics.keys() if k.endswith('/episodic_return')]
++                if len(ep_keys) > 0:
++                    ep_vals = []
++                    for k in ep_keys:
++                        tag = k.split('/')[0]
++                        denom = self.env_nums.get(tag, max(1, len(metrics[k])))
++                        ep_vals.append(float(np.sum(metrics[k]) / denom))
++                    mean_metrics["charts/avg_episode_return"] = float(np.mean(ep_vals))
++            except Exception:
++                pass
++            try:
++                # rollout/max_tile: max over all envs in this batch
++                tile_keys = [k for k in metrics.keys() if k.endswith('/max_tile')]
++                if len(tile_keys) > 0:
++                    tile_vals = []
++                    for k in tile_keys:
++                        tile_vals.extend(metrics[k])
++                    if len(tile_vals) > 0:
++                        mean_metrics["rollout/max_tile"] = int(np.max(tile_vals))
++            except Exception:
++                pass
+             for key, values in metrics.items():
+                 if not isinstance(values, list):
+                     continue
+diff --git a/ragen/llm_agent/es_manager.py b/ragen/llm_agent/es_manager.py
+index b87a2b3..1a49d3e 100644
+--- a/ragen/llm_agent/es_manager.py
++++ b/ragen/llm_agent/es_manager.py
+@@ -128,18 +128,40 @@ class EnvStateManager:
+         env_outputs: List[Dict]
+             {env_id: int, history: List[Dict][{state: str, actions: List[str], reward: float, info: Dict, llm_response: str, llm_raw_response: str, (Optional)images: List[PIL.Image.Image]}]}
+         """
++        # def _execute_actions(env, actions):
++        #     acc_reward, turn_info, turn_done = 0, {}, False
++        #     executed_actions = []
++        #     for action in actions:
++        #         _, reward, done, info = env.step(action)
++        #         acc_reward += reward
++        #         turn_info.update(info) # NOTE: currently use last info for multi-action
++        #         executed_actions.append(action)
++        #         if done:
++        #             turn_done = True
++        #             break
++            
++        #     return acc_reward, turn_info, turn_done, executed_actions
+         def _execute_actions(env, actions):
+-            acc_reward, turn_info, turn_done = 0, {}, False
++            acc_reward, turn_info, turn_done = 0.0, {}, False
++            raw_acc_reward = 0.0
+             executed_actions = []
+             for action in actions:
+                 _, reward, done, info = env.step(action)
+-                acc_reward += reward
++                acc_reward += float(reward)
++                try:
++                    raw_acc_reward += float(info.get('raw_reward', 0.0))
++                except Exception:
++                    pass
+                 turn_info.update(info) # NOTE: currently use last info for multi-action
+                 executed_actions.append(action)
+                 if done:
+                     turn_done = True
+                     break
+-            
++            # Overwrite per-turn raw_reward to reflect the sum across all executed actions in this turn
++            try:
++                turn_info['raw_reward'] = float(raw_acc_reward)
++            except Exception:
++                pass
+             return acc_reward, turn_info, turn_done, executed_actions
+ 
+         def _log_env_state(status, history, cur_obs, max_actions_per_traj, executed_actions, all_actions, acc_reward, turn_done, turn_info, env_input):
+@@ -198,6 +220,18 @@ class EnvStateManager:
+                 'success': float(status.terminated and (not status.truncated)),
+                 'num_actions': status.num_actions,
+             }
++            # Add episodic-level metrics
++            # try:
++            #     # Sum of per-turn rewards equals the episodic return (env-shaped reward)
++            #     env_metric['episodic_return'] = float(sum(status.rewards))
++            # except Exception:
++            #     pass
++            try:
++                # Final max tile on the board at the end of the rollout
++                import numpy as _np
++                env_metric['max_tile'] = int(_np.max(entry['env'].grid))
++            except Exception:
++                pass
+             custom_metric = {}
+             for turn in cache['history']:
+                 for k, v in turn.get('info', {}).items():
+@@ -212,6 +246,12 @@ class EnvStateManager:
+                             "Skipping non-numeric metric '%s' with value %r for env %s.",
+                             k, v, entry['tag']
+                         )
++            # Add episodic_return as the SUM of raw_reward across steps (align with CleanRL)
++            try:
++                if 'raw_reward' in custom_metric:
++                    env_metric['episodic_return'] = float(np.sum(custom_metric['raw_reward']))
++            except Exception:
++                pass
+             for k, v in custom_metric.items():
+                 # TODO: Move TURN_LVL_METRICS into the environment
+                 if "webshop" not in cache['tag'].lower() or ("webshop" in cache['tag'].lower() and k in TURN_LVL_METRICS):
+@@ -219,7 +259,12 @@ class EnvStateManager:
+                 else:
+                     env_metric['traj_sum/' + k] = np.sum(v)
+ 
+-
++            try:
++                if 'score' in custom_metric and len(custom_metric['score']) > 0:
++                    env_metric['final_score'] = float(custom_metric['score'][-1])
++            except Exception:
++                pass
++                
+             cache['history'][-1]['metrics'] = custom_metric
+             env_metric = {f"{entry['tag']}/{k}": v for k, v in env_metric.items()}
+             cache['metrics'] = env_metric
+diff --git a/requirements.txt b/requirements.txt
+index 2fd756e..9b521bf 100644
+--- a/requirements.txt
++++ b/requirements.txt
+@@ -7,7 +7,6 @@ accelerate
+ codetiming
+ datasets
+ dill
+-flash-attn==2.7.4.post1
+ hydra-core
+ numpy
+ pandas
+@@ -15,19 +14,19 @@ pybind11
+ ray>=2.10
+ tensordict>=0.8.0,<0.9.0
+ transformers
+-vllm==0.8.2
++vllm==0.8.5
+ wandb
+ gymnasium
+ gymnasium[toy-text]
+ 
+ pyarrow>=15.0.0
+ pylatexenc
+-torchdata
++# torchdata
+ debugpy
+ 
+ together
+ anthropic
+-faiss-cpu==1.11.0
++faiss-gpu
+ 
+ # This is optional, but needs to be installed with main requirements if you want to use webshop
+ -r external/webshop-minimal/requirements.txt
+diff --git a/scripts/setup_ragen.sh b/scripts/setup_ragen.sh
+index f9a7cd9..85a93c4 100644
+--- a/scripts/setup_ragen.sh
++++ b/scripts/setup_ragen.sh
+@@ -94,10 +94,10 @@ main() {
+         pip install torch==2.5.0 --index-url https://download.pytorch.org/whl/cu124
+         
+         print_step "Installing flash-attention..."
+-        pip3 install flash-attn==2.7.4.post1 --no-build-isolation
++        # pip3 install flash-attn==2.7.4.post1 --no-build-isolation
+     else
+         print_step "Installing PyTorch without CUDA support..."
+-        pip install torch==2.5.0
++        pip install torch==2.4.0
+     fi
+     
+     # Install remaining requirements
+@@ -137,8 +137,8 @@ main() {
+     conda install conda-forge::gdown
+     mkdir -p external/webshop-minimal/webshop_minimal/data/full
+     cd external/webshop-minimal/webshop_minimal/data/full
+-    gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
+-    gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
++    # gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
++    # gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
+     cd ../../../../..
+ 
+     echo -e "${GREEN}Installation completed successfully!${NC}"
+diff --git a/train_all.sh b/train_all.sh
+index 0157306..33035f5 100755
+--- a/train_all.sh
++++ b/train_all.sh
+@@ -6,246 +6,257 @@ USE_GRPO="algorithm.adv_estimator=grpo"
+ USE_PPO="algorithm.adv_estimator=gae" # by default.
+ USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+ 
++
++# python train.py --config-name _8_2048 system.CUDA_VISIBLE_DEVICES="'0,1,2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=game_2048 $USE_PPO $USE_BASE 
++
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-1 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-1.5B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'4,5'" trainer.project_name=ragen_latest_qwen_25_3b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-3B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-3 $USE_PPO $USE_BASE 
++
+ # Section 3.1&3.2 - General Observations
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozen_lake-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozen_lake-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=bandit-ppo-multitask $USE_PPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'7'" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-ppo-frommlp $USE_PPO $USE_BASE 
+ 
+-# Section 4.1 - Filtering and critic
+-# 0.25
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-
+-wait
+-
+-# 0.5
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-
+-# 0.75
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-
+-wait
+-
+-# Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=sokoban-ppo-box1-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-wait
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.project_name=ragen_latest_qwen_25_3b_it  trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_3B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-# Section 5.1 - Reasoning Helps Generalization
++# # Section 4.1 - Filtering and critic
++# # 0.25
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
+-    agent_proxy.enable_think=False
++# # 0.5
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++
++# # 0.75
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
+-    agent_proxy.enable_think=False
++# # Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+ 
++# wait
+ 
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
++# # Section 5.1 - Reasoning Helps Generalization
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
+ 
+-# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
++#     agent_proxy.enable_think=False
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader"
+ 
+-# COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
++#     agent_proxy.enable_think=False
+ 
+-wait
+ 
+-# Section 5.2 - what leads to better reasoning?
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
++# # SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
++# # COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
+ 
++# wait
+ 
+-wait
++# # Section 5.2 - what leads to better reasoning?
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
+ 
+-# Section 6.1 varying action count
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
++# wait
+ 
+-# section 6.2 Varying prompt diversity
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
+ 
+-wait 
++# wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
++# wait
+ 
++# # Section 6.1 varying action count
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
+ 
++# # section 6.2 Varying prompt diversity
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
+ 
++# wait 
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
+ 
+-# Extension: Training 7B reasoning model
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
+ 
+-# trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
++# # Extension: Training 7B reasoning model
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
+ 
++# python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
+ 
++# # trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_PPO="algorithm.adv_estimator=gae" # by default.
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
+-# enable_response_mask: False
+-# grpo_advantage_length_weight: True
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+ 
+-# extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
++# # USE_PPO="algorithm.adv_estimator=gae" # by default.
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
+ 
+-# extension: bi-level gae
+-python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
+-    system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
+-    model_path=Qwen/Qwen2.5-0.5B-Instruct \
+-    algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
+-    agent_proxy.use_turn_scores=True \
+-    actor_rollout_ref.rollout.tp_size_check=False
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
+ 
+-# extension: webshop
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # enable_response_mask: False
++# # grpo_advantage_length_weight: True
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: bi-level gae
++# python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
++#     system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
++#     model_path=Qwen/Qwen2.5-0.5B-Instruct \
++#     algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
++#     agent_proxy.use_turn_scores=True \
++#     actor_rollout_ref.rollout.tp_size_check=False
+ 
+ 
+-# normal:sokoban
+-# extension: sokoban
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # extension: webshop
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
+ 
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++
++# # normal:sokoban
++# # extension: sokoban
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
++
++
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
++
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+ 
+ 
+-python train.py \
+-    trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+
++# python train.py \
++#     trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+

wandb/run-20251218_183419-wvypkl8w/files/diff_8d73639b99b38265453f898b8d6af7d4af50d56e.patch

@@ -0,0 +1,1162 @@
+diff --git a/config/base.yaml b/config/base.yaml
+index a1659e4..9df6250 100644
+--- a/config/base.yaml
++++ b/config/base.yaml
+@@ -9,15 +9,16 @@ seed:
+   train: 10000
+   val: 123
+ 
+-micro_batch_size_per_gpu: 4
++micro_batch_size_per_gpu: 1
+ ppo_mini_batch_size: 32
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/Qwen/Qwen2.5-0.5B-Instruct
++# Qwen/Qwen2.5-0.5B-Instruct
+ enable_response_mask: True # Enabling response mask could improve stability of rollout/old_log_prob, as P(st|history) are no longer calculated in loss here. See https://docs.google.com/document/d/1bg7obeiKTExuHHBl5uOiSpec5uLDZ2Tgvxy6li5pHX4/edit?usp=sharing for more details.
+ grpo_advantage_length_weight: False # if you do not enable this and critic/advantage_estimator is GRPO, and the critic/advantages/mean is too low, then you can try enabling this to encourage reasoning and forbid collapse
+ 
+ lora:
+   rank: 0
+-  alpha: 64
++  alpha: 16
+   target_modules: all-linear
+ 
+ actor_rollout_ref:
+@@ -46,10 +47,10 @@ actor_rollout_ref:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu} # following micro_batch_size_per_gpu
+     tensor_model_parallel_size: 1
+-    max_model_len: 3600
++    max_model_len: 7200 #3600 why** 14400
+     prompt_length: 1 # useless. Just put it here
+     response_length: 400 # single-turn response length
+-    gpu_memory_utilization: 0.5
++    gpu_memory_utilization: 0.7
+     max_num_batched_tokens: 8192 # set only when enable_chunked_prefill is true
+     temperature: 1
+     rollout_filter_ratio: 0.25
+@@ -90,26 +91,28 @@ algorithm:
+     kl_coef: 0.000
+ 
+ trainer:
+-  project_name: ragen_latest
++  project_name: 
+   experiment_name: test
+   local_log_dir: "results/"
+-  save_freq: 100
++  save_freq: -1
+   total_training_steps: 200
+   validation_steps: 1 # validation instances = validation_steps * val_env_groups * group_size
+   val_before_train: True
+   n_gpus_per_node: 1
+   test_freq: 10
+   generations_to_log_to_wandb: 
++    train: 128
+     val: 20
+   logger: [ 'console', 'wandb' ]
+   max_actor_ckpt_to_keep: 1
+   max_critic_ckpt_to_keep: 1
++  default_local_dir: /mnt/general/wanghy/RAGEN/saves/
+ 
+ agent_proxy:
+   max_context_window: -1 # set a value > 0 to enable context window for long trajectory
+-  max_turn: 5
++  max_turn: 25 #25 why** 700
+   action_sep: "||"
+-  max_actions_per_turn: 2 # how many actions can be output at most in a single turn
++  max_actions_per_turn: 1 # how many actions can be output at most in a single turn
+   use_turn_scores: False # important to GAE when applying token-level rewards to token-level advantages. If False, will take the sum of scores as the reward for the last turn.
+   enable_think: True # False -> no think RL
+   reward_normalization:
+@@ -126,11 +129,11 @@ es_manager:
+       tags: ["CoordSokoban"]
+       n_groups: [8] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+   val:
+-    env_groups: 32
+-    group_size: 16 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
++    env_groups: 256
++    group_size: 1 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
+     env_configs:
+       tags: ["CoordSokoban"]
+-      n_groups: [32] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      n_groups: [256] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ ctx_manager:
+   generation: # go to vllm
+diff --git a/config/envs.yaml b/config/envs.yaml
+index 1002ac8..ba3ac6f 100644
+--- a/config/envs.yaml
++++ b/config/envs.yaml
+@@ -2,6 +2,7 @@ custom_envs:
+   SimpleSokoban:
+     env_type: sokoban
+     max_actions_per_traj: 10 # used in environment state manager to control the actual max actions executed per trajectory
++    # ORIGNAL env_instruction
+     env_instruction: |
+       You are solving the Sokoban puzzle. 
+       You are the player and you need to push all boxes to targets. 
+@@ -12,7 +13,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+ 
+   LargerSokoban:
+@@ -65,7 +66,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+       observation_format: "grid_coord"
+ 
+@@ -111,14 +112,14 @@ custom_envs:
+ 
+   FrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: "You are solving the FrozenLake puzzle. Forbid the whole and go to the target. You may move to the unintended direction due to the slippery ice. Example answer format: <think>To forbid the hole and go to the target, I should go left then go up.</think><answer>Left || Up</answer>"
+     max_tokens: 100
+     env_config: null
+ 
+   CoordFrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: |
+       You are solving the FrozenLake puzzle. The observation includes both a symbol grid and zero-indexed coordinates for the start, goal, player, and any holes.
+       Coordinates range from the top-left corner (0, 0) to the bottom-right corner (5, 5).
+@@ -195,4 +196,80 @@ custom_envs:
+     max_actions_per_traj: 30
+     env_instruction: "You are a Lean theorem prover. Given a Lean theorem statement, propose a sequence of tactics that completes the proof. Think step by step about which tactics to apply next. Provide tactics separated by '||', for example <answer>intro || simp || rfl</answer>."
+     max_tokens: 512
+-    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
+
++    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
++
++  
++  game_2048:
++    env_type: game_2048
++    max_actions_per_traj: 700
++    env_instruction: |
++      You are playing the 2048 game on a 4x4 grid. Merge equal tiles by sliding Up, Right, Down, or Left.
++      If a move is invalid (no tiles move), a small penalty is applied. Respond with a single action.
++      Example: <answer>Up</answer>
++    max_tokens: 8192
++    env_config: null
++
++  blackjack:
++    env_type: blackjack
++    max_actions_per_traj: 10
++    env_instruction: |
++      You are playing Blackjack against a dealer. The dealer must hit on 16 or less and stand on 17 or more.
++      Choose either Stick or Hit. Respond with a single action.
++      Example: <answer>Hit</answer>
++    max_tokens: 64
++    env_config: null
++
++  rubikscube:
++    env_type: rubikscube
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.
++      Available actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.
++      - Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).
++      - Modifiers: A letter alone means 90° clockwise (e.g., 'R'). A letter with prime (') means 90° counter-clockwise (e.g., "R'").
++      Respond with a sequence of actions separated by "||".
++      Example: <answer>U</answer>
++    max_tokens: 96
++    env_config:
++      scramble_depth: 1
++      max_steps: 20
++      render_mode: "text"
++  
++  SimpleSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 2x2 box contains the numbers 1-4 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format, for example: <answer>place 1 at row 2 col 3</answer> or <answer>1,2,3</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 4
++      difficulty: "easy"
++      render_format: "with_feedback"
++      show_conflicts: false
++      show_valid_numbers: false
++      max_steps: 20
++
++  MediumSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 30
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 3x3 box contains the numbers 1-9 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format: <answer>place 5 at row 2 col 3</answer> or <answer>2,3,5</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 9
++      difficulty: "medium"
++      render_format: "with_feedback"
++      show_conflicts: true
++      show_valid_numbers: true
++      max_steps: 81
++
+diff --git a/config/eval.yaml b/config/eval.yaml
+index e65f3a5..7775fd1 100644
+--- a/config/eval.yaml
++++ b/config/eval.yaml
+@@ -1,14 +1,15 @@
+ defaults:
+   - envs
+-
++enable_response_mask: True
+ system:
+-  CUDA_VISIBLE_DEVICES: "0"
++  CUDA_VISIBLE_DEVICES: "0,1,2,3,4,5,6,7"
+ 
+ seed:
+   train: 10000
+   val: 123
+ 
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/openai/gpt-oss-20b
++# /mnt/general/share/model/tyzhu/SPA-frozenlake-qwen2.5-1.5b-instruct
+ 
+ lora:
+   rank: 0
+@@ -24,9 +25,9 @@ actor_rollout_ref:
+   rollout:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu}
+-    tensor_model_parallel_size: 1
++    tensor_model_parallel_size: 8
+     dtype: bfloat16
+-    max_model_len: 3600
++    max_model_len: 7200
+     prompt_length: 1
+     response_length: 400
+     gpu_memory_utilization: 0.9
+@@ -35,6 +36,7 @@ actor_rollout_ref:
+     free_cache_engine: True
+     enable_chunked_prefill: False
+     disable_log_stats: False
++    do_sample: True
+     val_kwargs:
+       do_sample: True
+       temperature: 0.5
+@@ -44,9 +46,9 @@ actor_rollout_ref:
+ 
+ agent_proxy:
+   max_context_window: -1
+-  max_turn: 5
++  max_turn: 25
+   action_sep: "||"
+-  max_actions_per_turn: 2
++  max_actions_per_turn: 1
+   use_turn_scores: False
+   enable_think: True
+   reward_normalization:
+@@ -59,13 +61,13 @@ es_manager:
+     env_groups: 8
+     group_size: 16
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [8]
+   val:
+     env_groups: 32
+-    group_size: 16
++    group_size: 128
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [32]
+ 
+ ctx_manager:
+diff --git a/config/evaluate_api_llm.yaml b/config/evaluate_api_llm.yaml
+index b21c95e..e2e6761 100644
+--- a/config/evaluate_api_llm.yaml
++++ b/config/evaluate_api_llm.yaml
+@@ -1,8 +1,11 @@
++#export OPENAI_BASE_URL="https://api.ohmygpt.com/v1"
++#export OPENAI_API_KEY="sk-o4sMxBkN5BB100C4D4a3T3BlBkFJF7791CA39EA14ca98041"
++#python -m ragen.eval_api hydra.searchpath='[file://./verl/verl/trainer/config]'
+ defaults:
+   - base # this is a symbolic link to the verl/verl/trainer/config/ppo_trainer.yaml file
+ 
+ model_config:
+-  model_name: gpt-4o # should be registered in model_info
++  model_name: ark-deepseek-v3-250324 # should be registered in model_info
+   max_concurrency: 16
+ 
+ model_info:
+@@ -24,9 +27,9 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+-  gpt-4o:
++  gpt-4o-mini:
+     provider_name: openai
+-    model_name: gpt-4o
++    model_name: gpt-4o-mini
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+@@ -36,21 +39,42 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  ark-deepseek-v3-250324:
++    provider_name: openai
++    model_name: ark-deepseek-v3-250324
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
+   deepseek-v3:
+     provider_name: deepseek
+     model_name: deepseek-chat
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  glm-4.6:
++    provider_name: openai
++    model_name: glm-4.6
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
++  TA/openai/gpt-oss-120b:
++    provider_name: openai
++    model_name: TA/openai/gpt-oss-120b
++    generation_kwargs:
++      temperature: 0
++      max_tokens: 8192
++      # max_retries: 5
+ 
+-
+-
++agent_proxy:
++  max_turn: 5
+ es_manager:
+   val:
+-    env_groups: 256
++    env_groups: 128
+     group_size: 1 # should be set to 1 because val temperature is set to 0 and same prompt leads to same output
+     env_configs:
+-      tags: ["CoordSokoban"]
+-      n_groups: [256] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      tags: ["rubikscube"]
++      n_groups: [128] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ 
++rollout:
++  max_model_len: 7200
+
+Submodule external/webshop-minimal contains modified content
+diff --git a/external/webshop-minimal/requirements.txt b/external/webshop-minimal/requirements.txt
+index 5a1b04f..238ed5a 100644
+--- a/external/webshop-minimal/requirements.txt
++++ b/external/webshop-minimal/requirements.txt
+@@ -4,7 +4,7 @@ flask
+ html2text
+ rank_bm25
+ pyserini
+-faiss-cpu
++faiss-gpu
+ thefuzz
+ gdown
+ spacy
+diff --git a/ragen/env/__init__.py b/ragen/env/__init__.py
+index b0f3461..bc6838e 100644
+--- a/ragen/env/__init__.py
++++ b/ragen/env/__init__.py
+@@ -12,6 +12,14 @@ from .metamathqa.env import MetaMathQAEnv
+ from .metamathqa.config import MetaMathQAEnvConfig
+ from .lean.config import LeanEnvConfig
+ from .lean.env import LeanEnv
++from .game_2048.config import Game2048EnvConfig
++from .game_2048.env import Game2048Env
++from .blackjack.config import BlackjackEnvConfig
++from .blackjack.env import BlackjackEnv
++from .rubikscube.config import RubiksCube2x2Config
++from .rubikscube.env import RubiksCube2x2Env
++from .sudoku.config import SudokuEnvConfig
++from .sudoku.env import SudokuEnv
+ 
+ 
+ REGISTERED_ENVS = {
+@@ -22,6 +30,10 @@ REGISTERED_ENVS = {
+     # 'alfworld': AlfredTXTEnv,
+     'metamathqa': MetaMathQAEnv,
+     'lean': LeanEnv,
++    'game_2048': Game2048Env,
++    'blackjack': BlackjackEnv,
++    'rubikscube': RubiksCube2x2Env,
++    'sudoku': SudokuEnv,
+ }
+ 
+ REGISTERED_ENV_CONFIGS = {
+@@ -32,6 +44,10 @@ REGISTERED_ENV_CONFIGS = {
+     # 'alfworld': AlfredEnvConfig,
+     'metamathqa': MetaMathQAEnvConfig,
+     'lean': LeanEnvConfig,
++    'game_2048': Game2048EnvConfig,
++    'blackjack': BlackjackEnvConfig,
++    'rubikscube': RubiksCube2x2Config,
++    'sudoku': SudokuEnvConfig,
+ }
+ 
+ try:
+diff --git a/ragen/env/frozen_lake/config.py b/ragen/env/frozen_lake/config.py
+index de054f4..9950c34 100644
+--- a/ragen/env/frozen_lake/config.py
++++ b/ragen/env/frozen_lake/config.py
+@@ -8,7 +8,7 @@ class FrozenLakeEnvConfig:
+     size: int = 4
+     p: float = 0.9
+     success_rate: float = 0.8
+-    is_slippery: bool = True
++    is_slippery: bool = False
+     map_seed: Optional[int] = None
+     render_mode: str = "text"
+     observation_format: str = "grid"
+diff --git a/ragen/env/frozen_lake/env.py b/ragen/env/frozen_lake/env.py
+index 9ee3add..eef11d8 100644
+--- a/ragen/env/frozen_lake/env.py
++++ b/ragen/env/frozen_lake/env.py
+@@ -13,6 +13,7 @@ from ragen.env.base import BaseDiscreteActionEnv
+ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+     def __init__(self, config: FrozenLakeEnvConfig = FrozenLakeEnvConfig()):
+         # Using mappings directly from config
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.GRID_LOOKUP = config.grid_lookup
+         self.ACTION_LOOKUP = config.action_lookup
+@@ -95,6 +96,7 @@ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+ if __name__ == "__main__":
+     import matplotlib.pyplot as plt
+     config = FrozenLakeEnvConfig(size=4, is_slippery=True)
++    # import pdb;pdb.set_trace()
+     env = FrozenLakeEnv(config)
+     print(env.reset())
+     while True:
+diff --git a/ragen/env/sokoban/env.py b/ragen/env/sokoban/env.py
+index 17cd636..6922b54 100644
+--- a/ragen/env/sokoban/env.py
++++ b/ragen/env/sokoban/env.py
+@@ -33,6 +33,7 @@ class SokobanEnv(BaseDiscreteActionEnv, GymSokobanEnv):
+     def reset(self, seed=None, mode=None):
+         try:
+             with all_seed(seed):
++                # import pdb;pdb.set_trace()
+                 self.room_fixed, self.room_state, self.box_mapping, action_sequence = generate_room(
+                     dim=self.dim_room,
+                     num_steps=self.num_gen_steps,
+diff --git a/ragen/llm_agent/agent_proxy.py b/ragen/llm_agent/agent_proxy.py
+index c25b50d..f6c4088 100644
+--- a/ragen/llm_agent/agent_proxy.py
++++ b/ragen/llm_agent/agent_proxy.py
+@@ -287,6 +287,7 @@ def main(config):
+ 	proxy = LLMAgentProxy(config, actor_wg, tokenizer)
+ 	import time
+ 	start_time = time.time()
++	# import pdb;pdb.set_trace()
+ 	rollouts = proxy.rollout(
+ 		DataProto(
+ 			batch=None,
+diff --git a/ragen/llm_agent/base_llm.py b/ragen/llm_agent/base_llm.py
+index 358a2eb..b6f7797 100644
+--- a/ragen/llm_agent/base_llm.py
++++ b/ragen/llm_agent/base_llm.py
+@@ -33,6 +33,7 @@ class OpenAIProvider(LLMProvider):
+             raise ValueError("OpenAI API key not provided and not found in environment variables")
+         
+         self.client = AsyncOpenAI(api_key=self.api_key)
++        # import pdb;pdb.set_trace()
+     
+     async def generate(self, messages: List[Dict[str, str]], **kwargs) -> LLMResponse:
+         if "o1-mini" in self.model_name:
+@@ -46,10 +47,7 @@ class OpenAIProvider(LLMProvider):
+         )
+         if response.choices[0].finish_reason in ['length', 'content_filter']:
+             raise ValueError("Content filtered or length exceeded")
+-        return LLMResponse(
+-            content=response.choices[0].message.content,
+-            model_name=response.model
+-        )
++        return LLMResponse(content=response.choices[0].message.content,model_name=response.model)
+ 
+ class DeepSeekProvider(LLMProvider):
+     """DeepSeek API provider implementation"""
+@@ -199,7 +197,7 @@ class ConcurrentLLM:
+         
+         # Queue to store unfinished or failed tasks
+         current_batch = messages_list.copy()
+-        max_retries = kwargs.get("max_retries", 100)
++        max_retries = kwargs.get("max_retries", 10)
+         retry_count = 0
+         
+         while current_batch and retry_count < max_retries:
+diff --git a/ragen/llm_agent/ctx_manager.py b/ragen/llm_agent/ctx_manager.py
+index 905247a..af20a01 100644
+--- a/ragen/llm_agent/ctx_manager.py
++++ b/ragen/llm_agent/ctx_manager.py
+@@ -88,6 +88,7 @@ class ContextManager:
+         Initialize the ContextManager.
+         Processor is used to process the image data.
+         """
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.tokenizer = tokenizer
+         self.processor = processor
+@@ -311,7 +312,7 @@ class ContextManager:
+ 
+             llm_input_texts.append(text_with_prompt)
+             messages_list.append(messages)
+-
++        # import pdb;pdb.set_trace()
+         inputs = self.tokenizer(llm_input_texts, return_tensors="pt", padding=True, padding_side="left", truncation=False) # We have truncated previously, truncation in tokenizer may cause issues.
+         input_ids, attention_mask = inputs.input_ids, inputs.attention_mask
+         position_ids = (attention_mask.cumsum(dim=-1) - 1).clamp(min=0)
+@@ -353,6 +354,30 @@ class ContextManager:
+                 key: np.sum(value) / self.env_nums[key.split("/")[0]]
+                 for key, value in metrics.items()
+             }
++            # Derived metrics for wandb logging
++            try:
++                # charts/avg_episode_return: average across all 2048 env groups
++                ep_keys = [k for k in metrics.keys() if k.endswith('/episodic_return')]
++                if len(ep_keys) > 0:
++                    ep_vals = []
++                    for k in ep_keys:
++                        tag = k.split('/')[0]
++                        denom = self.env_nums.get(tag, max(1, len(metrics[k])))
++                        ep_vals.append(float(np.sum(metrics[k]) / denom))
++                    mean_metrics["charts/avg_episode_return"] = float(np.mean(ep_vals))
++            except Exception:
++                pass
++            try:
++                # rollout/max_tile: max over all envs in this batch
++                tile_keys = [k for k in metrics.keys() if k.endswith('/max_tile')]
++                if len(tile_keys) > 0:
++                    tile_vals = []
++                    for k in tile_keys:
++                        tile_vals.extend(metrics[k])
++                    if len(tile_vals) > 0:
++                        mean_metrics["rollout/max_tile"] = int(np.max(tile_vals))
++            except Exception:
++                pass
+             for key, values in metrics.items():
+                 if not isinstance(values, list):
+                     continue
+diff --git a/ragen/llm_agent/es_manager.py b/ragen/llm_agent/es_manager.py
+index b87a2b3..1a49d3e 100644
+--- a/ragen/llm_agent/es_manager.py
++++ b/ragen/llm_agent/es_manager.py
+@@ -128,18 +128,40 @@ class EnvStateManager:
+         env_outputs: List[Dict]
+             {env_id: int, history: List[Dict][{state: str, actions: List[str], reward: float, info: Dict, llm_response: str, llm_raw_response: str, (Optional)images: List[PIL.Image.Image]}]}
+         """
++        # def _execute_actions(env, actions):
++        #     acc_reward, turn_info, turn_done = 0, {}, False
++        #     executed_actions = []
++        #     for action in actions:
++        #         _, reward, done, info = env.step(action)
++        #         acc_reward += reward
++        #         turn_info.update(info) # NOTE: currently use last info for multi-action
++        #         executed_actions.append(action)
++        #         if done:
++        #             turn_done = True
++        #             break
++            
++        #     return acc_reward, turn_info, turn_done, executed_actions
+         def _execute_actions(env, actions):
+-            acc_reward, turn_info, turn_done = 0, {}, False
++            acc_reward, turn_info, turn_done = 0.0, {}, False
++            raw_acc_reward = 0.0
+             executed_actions = []
+             for action in actions:
+                 _, reward, done, info = env.step(action)
+-                acc_reward += reward
++                acc_reward += float(reward)
++                try:
++                    raw_acc_reward += float(info.get('raw_reward', 0.0))
++                except Exception:
++                    pass
+                 turn_info.update(info) # NOTE: currently use last info for multi-action
+                 executed_actions.append(action)
+                 if done:
+                     turn_done = True
+                     break
+-            
++            # Overwrite per-turn raw_reward to reflect the sum across all executed actions in this turn
++            try:
++                turn_info['raw_reward'] = float(raw_acc_reward)
++            except Exception:
++                pass
+             return acc_reward, turn_info, turn_done, executed_actions
+ 
+         def _log_env_state(status, history, cur_obs, max_actions_per_traj, executed_actions, all_actions, acc_reward, turn_done, turn_info, env_input):
+@@ -198,6 +220,18 @@ class EnvStateManager:
+                 'success': float(status.terminated and (not status.truncated)),
+                 'num_actions': status.num_actions,
+             }
++            # Add episodic-level metrics
++            # try:
++            #     # Sum of per-turn rewards equals the episodic return (env-shaped reward)
++            #     env_metric['episodic_return'] = float(sum(status.rewards))
++            # except Exception:
++            #     pass
++            try:
++                # Final max tile on the board at the end of the rollout
++                import numpy as _np
++                env_metric['max_tile'] = int(_np.max(entry['env'].grid))
++            except Exception:
++                pass
+             custom_metric = {}
+             for turn in cache['history']:
+                 for k, v in turn.get('info', {}).items():
+@@ -212,6 +246,12 @@ class EnvStateManager:
+                             "Skipping non-numeric metric '%s' with value %r for env %s.",
+                             k, v, entry['tag']
+                         )
++            # Add episodic_return as the SUM of raw_reward across steps (align with CleanRL)
++            try:
++                if 'raw_reward' in custom_metric:
++                    env_metric['episodic_return'] = float(np.sum(custom_metric['raw_reward']))
++            except Exception:
++                pass
+             for k, v in custom_metric.items():
+                 # TODO: Move TURN_LVL_METRICS into the environment
+                 if "webshop" not in cache['tag'].lower() or ("webshop" in cache['tag'].lower() and k in TURN_LVL_METRICS):
+@@ -219,7 +259,12 @@ class EnvStateManager:
+                 else:
+                     env_metric['traj_sum/' + k] = np.sum(v)
+ 
+-
++            try:
++                if 'score' in custom_metric and len(custom_metric['score']) > 0:
++                    env_metric['final_score'] = float(custom_metric['score'][-1])
++            except Exception:
++                pass
++                
+             cache['history'][-1]['metrics'] = custom_metric
+             env_metric = {f"{entry['tag']}/{k}": v for k, v in env_metric.items()}
+             cache['metrics'] = env_metric
+diff --git a/requirements.txt b/requirements.txt
+index 2fd756e..9b521bf 100644
+--- a/requirements.txt
++++ b/requirements.txt
+@@ -7,7 +7,6 @@ accelerate
+ codetiming
+ datasets
+ dill
+-flash-attn==2.7.4.post1
+ hydra-core
+ numpy
+ pandas
+@@ -15,19 +14,19 @@ pybind11
+ ray>=2.10
+ tensordict>=0.8.0,<0.9.0
+ transformers
+-vllm==0.8.2
++vllm==0.8.5
+ wandb
+ gymnasium
+ gymnasium[toy-text]
+ 
+ pyarrow>=15.0.0
+ pylatexenc
+-torchdata
++# torchdata
+ debugpy
+ 
+ together
+ anthropic
+-faiss-cpu==1.11.0
++faiss-gpu
+ 
+ # This is optional, but needs to be installed with main requirements if you want to use webshop
+ -r external/webshop-minimal/requirements.txt
+diff --git a/scripts/setup_ragen.sh b/scripts/setup_ragen.sh
+index f9a7cd9..85a93c4 100644
+--- a/scripts/setup_ragen.sh
++++ b/scripts/setup_ragen.sh
+@@ -94,10 +94,10 @@ main() {
+         pip install torch==2.5.0 --index-url https://download.pytorch.org/whl/cu124
+         
+         print_step "Installing flash-attention..."
+-        pip3 install flash-attn==2.7.4.post1 --no-build-isolation
++        # pip3 install flash-attn==2.7.4.post1 --no-build-isolation
+     else
+         print_step "Installing PyTorch without CUDA support..."
+-        pip install torch==2.5.0
++        pip install torch==2.4.0
+     fi
+     
+     # Install remaining requirements
+@@ -137,8 +137,8 @@ main() {
+     conda install conda-forge::gdown
+     mkdir -p external/webshop-minimal/webshop_minimal/data/full
+     cd external/webshop-minimal/webshop_minimal/data/full
+-    gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
+-    gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
++    # gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
++    # gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
+     cd ../../../../..
+ 
+     echo -e "${GREEN}Installation completed successfully!${NC}"
+diff --git a/train_all.sh b/train_all.sh
+index 0157306..33035f5 100755
+--- a/train_all.sh
++++ b/train_all.sh
+@@ -6,246 +6,257 @@ USE_GRPO="algorithm.adv_estimator=grpo"
+ USE_PPO="algorithm.adv_estimator=gae" # by default.
+ USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+ 
++
++# python train.py --config-name _8_2048 system.CUDA_VISIBLE_DEVICES="'0,1,2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=game_2048 $USE_PPO $USE_BASE 
++
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-1 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-1.5B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'4,5'" trainer.project_name=ragen_latest_qwen_25_3b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-3B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-3 $USE_PPO $USE_BASE 
++
+ # Section 3.1&3.2 - General Observations
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozen_lake-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozen_lake-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=bandit-ppo-multitask $USE_PPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'7'" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-ppo-frommlp $USE_PPO $USE_BASE 
+ 
+-# Section 4.1 - Filtering and critic
+-# 0.25
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-
+-wait
+-
+-# 0.5
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-
+-# 0.75
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-
+-wait
+-
+-# Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=sokoban-ppo-box1-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-wait
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.project_name=ragen_latest_qwen_25_3b_it  trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_3B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-# Section 5.1 - Reasoning Helps Generalization
++# # Section 4.1 - Filtering and critic
++# # 0.25
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
+-    agent_proxy.enable_think=False
++# # 0.5
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++
++# # 0.75
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
+-    agent_proxy.enable_think=False
++# # Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+ 
++# wait
+ 
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
++# # Section 5.1 - Reasoning Helps Generalization
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
+ 
+-# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
++#     agent_proxy.enable_think=False
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader"
+ 
+-# COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
++#     agent_proxy.enable_think=False
+ 
+-wait
+ 
+-# Section 5.2 - what leads to better reasoning?
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
++# # SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
++# # COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
+ 
++# wait
+ 
+-wait
++# # Section 5.2 - what leads to better reasoning?
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
+ 
+-# Section 6.1 varying action count
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
++# wait
+ 
+-# section 6.2 Varying prompt diversity
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
+ 
+-wait 
++# wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
++# wait
+ 
++# # Section 6.1 varying action count
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
+ 
++# # section 6.2 Varying prompt diversity
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
+ 
++# wait 
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
+ 
+-# Extension: Training 7B reasoning model
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
+ 
+-# trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
++# # Extension: Training 7B reasoning model
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
+ 
++# python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
+ 
++# # trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_PPO="algorithm.adv_estimator=gae" # by default.
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
+-# enable_response_mask: False
+-# grpo_advantage_length_weight: True
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+ 
+-# extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
++# # USE_PPO="algorithm.adv_estimator=gae" # by default.
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
+ 
+-# extension: bi-level gae
+-python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
+-    system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
+-    model_path=Qwen/Qwen2.5-0.5B-Instruct \
+-    algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
+-    agent_proxy.use_turn_scores=True \
+-    actor_rollout_ref.rollout.tp_size_check=False
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
+ 
+-# extension: webshop
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # enable_response_mask: False
++# # grpo_advantage_length_weight: True
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: bi-level gae
++# python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
++#     system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
++#     model_path=Qwen/Qwen2.5-0.5B-Instruct \
++#     algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
++#     agent_proxy.use_turn_scores=True \
++#     actor_rollout_ref.rollout.tp_size_check=False
+ 
+ 
+-# normal:sokoban
+-# extension: sokoban
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # extension: webshop
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
+ 
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++
++# # normal:sokoban
++# # extension: sokoban
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
++
++
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
++
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+ 
+ 
+-python train.py \
+-    trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+
++# python train.py \
++#     trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+

wandb/run-20251218_183419-wvypkl8w/files/requirements.txt

@@ -0,0 +1,315 @@
+colorama==0.4.6
+psutil==7.1.2
+ragen==0.1
+ragen==0.1
+setuptools==80.9.0
+wheel==0.45.1
+pip==25.3
+ragen==0.1
+pytz==2025.2
+nvidia-cusparselt-cu12==0.6.2
+mpmath==1.3.0
+typing_extensions==4.15.0
+sympy==1.13.1
+pylatexenc==2.10
+nvidia-cusolver-cu12==11.6.1.9
+numpy==1.26.4
+Markdown==3.10
+absl-py==2.3.1
+pyasn1_modules==0.4.2
+Werkzeug==3.1.3
+networkx==3.3
+MarkupSafe==2.1.5
+fsspec==2025.9.0
+filelock==3.19.1
+nvidia-cudnn-cu12==9.1.0.70
+Jinja2==3.1.6
+tensorboard==2.20.0
+triton==3.2.0
+pillow==11.3.0
+xxhash==3.6.0
+nvidia-nvtx-cu12==12.4.127
+nvidia-nvjitlink-cu12==12.4.127
+nvidia-nccl-cu12==2.21.5
+nvidia-curand-cu12==10.3.5.147
+wrapt==2.0.1
+nvidia-cufft-cu12==11.2.1.3
+pyjnius==1.7.0
+nvidia-cuda-runtime-cu12==12.4.127
+nvidia-cuda-nvrtc-cu12==12.4.127
+websockets==15.0.1
+nvidia-cuda-cupti-cu12==12.4.127
+zipp==3.23.0
+nvidia-cublas-cu12==12.4.5.8
+nvidia-cusparse-cu12==12.3.1.170
+flash_attn==2.8.3
+rsa==4.9.1
+einops==0.8.1
+py-cpuinfo==9.0.0
+pure_eval==0.2.3
+ptyprocess==0.7.0
+gym-notices==0.1.0
+flatbuffers==25.9.23
+fastrlock==0.8.3
+Farama-Notifications==0.0.4
+faiss-gpu==1.7.2
+antlr4-python3-runtime==4.9.3
+wcwidth==0.2.14
+wasabi==1.1.3
+uvloop==0.22.1
+urllib3==2.5.0
+tzdata==2025.2
+typing-inspection==0.4.2
+traitlets==5.14.3
+tqdm==4.67.1
+tomli==2.3.0
+threadpoolctl==3.6.0
+tenacity==9.1.2
+tabulate==0.9.0
+spacy-loggers==1.0.5
+spacy-legacy==3.0.12
+soupsieve==2.8
+sniffio==1.3.1
+smmap==5.0.2
+six==1.17.0
+shellingham==1.5.4
+sentencepiece==0.2.1
+scipy==1.15.3
+safetensors==0.6.2
+rpds-py==0.29.0
+rignore==0.7.6
+regex==2025.11.3
+RapidFuzz==3.14.3
+rank-bm25==0.2.2
+pyzmq==27.1.0
+PyYAML==6.0.3
+pytokens==0.3.0
+python-multipart==0.0.20
+python-json-logger==4.0.0
+python-dotenv==1.2.1
+PySocks==1.7.1
+pyparsing==3.2.5
+PyJWT==2.10.1
+Pygments==2.19.2
+pygame==2.6.1
+pydantic_core==2.41.5
+pycparser==2.23
+pycountry==24.6.1
+pybind11==3.0.1
+pyarrow==22.0.0
+psutil==7.1.3
+protobuf==4.25.8
+propcache==0.4.1
+prometheus_client==0.23.1
+platformdirs==4.5.0
+pexpect==4.9.0
+pathspec==0.12.1
+partial-json-parser==0.2.1.1.post6
+parso==0.8.5
+packaging==25.0
+orjson==3.11.4
+opentelemetry-semantic-conventions-ai==0.4.13
+opencv-python-headless==4.12.0.88
+ninja==1.13.0
+nest-asyncio==1.6.0
+mypy_extensions==1.1.0
+murmurhash==1.0.15
+multidict==6.7.0
+msgspec==0.19.0
+msgpack==1.1.2
+mdurl==0.1.2
+loguru==0.7.3
+llvmlite==0.44.0
+llguidance==0.7.30
+lark==1.2.2
+kiwisolver==1.4.9
+joblib==1.5.2
+jiter==0.12.0
+itsdangerous==2.2.0
+interegular==0.3.3
+ImageIO==2.37.2
+idna==3.11
+humanfriendly==10.0
+httpx-sse==0.4.3
+httptools==0.7.1
+html2text==2025.4.15
+hf-xet==1.2.0
+h11==0.16.0
+grpcio==1.76.0
+frozenlist==1.8.0
+fonttools==4.60.1
+executing==2.2.1
+exceptiongroup==1.3.0
+eval_type_backport==0.2.2
+docstring_parser==0.17.0
+dnspython==2.8.0
+distro==1.9.0
+diskcache==5.6.3
+dill==0.4.0
+decorator==5.2.1
+debugpy==1.8.17
+Cython==3.2.1
+cymem==2.0.13
+cycler==0.12.1
+cupy-cuda12x==13.6.0
+contourpy==1.3.2
+colorama==0.4.6
+codetiming==1.4.0
+cloudpickle==3.1.2
+cloudpathlib==0.23.0
+click==8.3.1
+charset-normalizer==3.4.4
+certifi==2025.11.12
+catalogue==2.0.10
+cachetools==6.2.2
+blis==1.3.2
+blinker==1.9.0
+blake3==1.0.8
+attrs==25.4.0
+async-timeout==5.0.1
+asttokens==3.0.1
+astor==0.8.1
+annotated-types==0.7.0
+annotated-doc==0.0.4
+airportsdata==20250909
+aiohappyeyeballs==2.6.1
+yarl==1.22.0
+uvicorn==0.38.0
+typer-slim==0.20.0
+thefuzz==0.22.1
+stack-data==0.6.3
+srsly==2.5.1
+smart_open==7.5.0
+sentry-sdk==2.44.0
+scikit-learn==1.7.2
+requests==2.32.5
+referencing==0.37.0
+python-dateutil==2.9.0.post0
+pydantic==2.12.4
+prompt_toolkit==3.0.52
+preshed==3.0.11
+opentelemetry-proto==1.26.0
+omegaconf==2.3.0
+numba==0.61.2
+nltk==3.9.2
+multiprocess==0.70.18
+matplotlib-inline==0.2.1
+markdown-it-py==4.0.0
+jedi==0.19.2
+importlib_metadata==8.0.0
+httpcore==1.0.9
+gymnasium==1.2.2
+gym==0.26.2
+googleapis-common-protos==1.72.0
+gitdb==4.0.12
+gguf==0.17.1
+Flask==3.1.2
+faiss-cpu==1.13.0
+email-validator==2.3.0
+depyf==0.18.0
+Deprecated==1.3.1
+coloredlogs==15.0.1
+cffi==2.0.0
+black==25.11.0
+beautifulsoup4==4.14.2
+anyio==4.11.0
+aiosignal==1.4.0
+watchfiles==1.1.1
+tiktoken==0.12.0
+starlette==0.49.3
+sse-starlette==3.0.3
+rich==14.2.0
+pydantic-settings==2.12.0
+pydantic-extra-types==2.10.6
+pandas==2.3.3
+opentelemetry-exporter-otlp-proto-common==1.26.0
+onnxruntime==1.23.2
+matplotlib==3.10.7
+lm-format-enforcer==0.10.12
+jsonschema-specifications==2025.9.1
+ipython==8.37.0
+hydra-core==1.3.2
+huggingface-hub==0.36.0
+httpx==0.28.1
+gym-sokoban==0.0.6
+GitPython==3.1.45
+cryptography==46.0.3
+confection==0.1.5
+cleantext==1.1.4
+aiohttp==3.13.2
+weasel==0.4.3
+wandb==0.23.0
+typer==0.19.2
+virtualenv==20.35.4
+pyasn1==0.6.1
+torch==2.6.0
+tokenizers==0.22.1
+thinc==8.3.9
+rich-toolkit==0.15.1
+prometheus-fastapi-instrumentator==7.1.0
+openai==2.8.0
+jsonschema==4.25.1
+gdown==5.2.0
+fastapi==0.121.2
+anthropic==0.73.0
+xformers==0.0.29.post2
+transformers==4.57.1
+proto-plus==1.26.1
+torchvision==0.21.0
+torchdata==0.11.0
+torchaudio==2.6.0
+together==1.5.30
+tensordict==0.8.3
+spacy==3.8.9
+ray==2.51.1
+outlines_core==0.1.26
+mistral_common==1.8.5
+mcp==1.21.1
+fastapi-cloud-cli==0.3.1
+fastapi-cli==0.0.16
+datasets==4.4.1
+accelerate==1.11.0
+xgrammar==0.1.18
+pyserini==1.2.0
+peft==0.18.0
+outlines==0.1.11
+opentelemetry-exporter-otlp-proto-http==1.26.0
+opentelemetry-exporter-otlp-proto-grpc==1.26.0
+kimina-client==0.2.1
+compressed-tensors==0.9.3
+opentelemetry-exporter-otlp==1.26.0
+vllm==0.8.5
+py-spy==0.4.1
+opencensus-context==0.1.3
+distlib==0.4.0
+colorful==0.5.8
+tensorboard-data-server==0.7.2
+google-auth==2.43.0
+google-api-core==2.28.1
+aiohttp-cors==0.8.1
+opentelemetry-exporter-prometheus==0.59b0
+opencensus==0.11.4
+verl==0.5.0.dev0
+opentelemetry-api==1.38.0
+opentelemetry-semantic-conventions==0.59b0
+opentelemetry-sdk==1.38.0
+typeguard==4.4.4
+shtab==1.8.0
+tyro==0.9.35
+verl==0.5.0.dev0
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2

wandb/run-20251218_183419-wvypkl8w/files/wandb-metadata.json

@@ -0,0 +1,92 @@
+{
+  "os": "Linux-5.14.0-284.25.1.el9_2.x86_64-x86_64-with-glibc2.35",
+  "python": "CPython 3.10.19",
+  "startedAt": "2025-12-18T10:34:19.463187Z",
+  "args": [
+    "--track"
+  ],
+  "program": "/mnt/general/wanghy/RAGEN/cleanrl/cleanrl/ppo_rubikscube.py",
+  "codePath": "cleanrl/cleanrl/ppo_rubikscube.py",
+  "codePathLocal": "cleanrl/cleanrl/ppo_rubikscube.py",
+  "git": {
+    "remote": "https://github.com/mll-lab-nu/RAGEN.git",
+    "commit": "8d73639b99b38265453f898b8d6af7d4af50d56e"
+  },
+  "email": "haoyu-wa22@mails.tsinghua.edu.cn",
+  "root": "/mnt/general/wanghy/RAGEN",
+  "host": "pt-4d654cf4576f4d23ad3d3919f12932fe-worker-0",
+  "executable": "/root/local/miniconda3/envs/ragen/bin/python",
+  "cpu_count": 64,
+  "cpu_count_logical": 128,
+  "gpu": "NVIDIA H100 80GB HBM3",
+  "gpu_count": 8,
+  "disk": {
+    "/": {
+      "total": "5153960755200",
+      "used": "447541800960"
+    }
+  },
+  "memory": {
+    "total": "2163642122240"
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-35e2d43d-4067-82ce-90d4-def9e389bf28"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-af4135e3-88f2-e9ac-518d-502c75a85429"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-d7fdeeba-fe9b-ec03-d9f7-6724fe4266b5"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-ccc4f668-3882-5a8e-2c07-c5cd08f6f666"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-7b73c0cf-d3d5-e10c-7176-a43be1e41001"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-81b58d94-5d1f-8ec2-f9d2-fd56172ed177"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-03e8bc66-3b44-6794-49fd-5392fbdda6d1"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-b7cf0ec6-7c29-1179-dceb-09565da51890"
+    }
+  ],
+  "cudaVersion": "12.4",
+  "writerId": "bay0zucrvgh0budvh5sfo6odu02gecuh"
+}

wandb/run-20251218_183419-wvypkl8w/files/wandb-summary.json

@@ -0,0 +1 @@
+{"_timestamp":1.7660543920527563e+09,"train/old_approx_kl":8.68411734700203e-06,"losses/explained_variance":0.5622017085552216,"losses/policy_loss":-3.0249357223510742e-05,"train/clipfrac":0,"eval/std_return":0.4994821693514194,"train/policy_loss":-3.0249357223510742e-05,"eval/episodes":4000,"rollout/ep_rew_mean":0,"train/value_loss":0.029777294024825096,"eval/avg_return":0.52275,"global_step":999424,"train/learning_rate":2.561475409836034e-07,"charts/SPS":3145,"charts/progress":100,"rollout/ep_len_mean":20,"charts/avg_value":0.21519102156162262,"charts/avg_reward":0.0439453125,"rollout/success_rate":0,"train/success_rate":0.5421686746987951,"train/approx_kl":4.7730281949043274e-08,"losses/value_loss":0.029777294024825096,"train/entropy":0.8965780735015869,"_runtime":331.418744783,"_step":999424,"charts/train_success_rate":0.5421686746987951,"_wandb":{"runtime":331},"perf/SPS":3145,"eval/success_rate":0.52275}

wandb/run-20251218_183419-wvypkl8w/run-wvypkl8w.wandb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86f0fd4887e69f34f8ff351b143fb12ef67fe287e45a910a42d951c480b15664
+size 26717484

wandb/run-20251218_184253-g8kppnjh/files/code/cleanrl/cleanrl/ppo_rubikscube.py

@@ -0,0 +1,517 @@
+# PPO with small MLP for RAGEN Rubik's Cube 2x2 using the existing env (no env edits)
+import os
+import random
+import time
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Tuple, Dict, Any, List
+import json
+import re
+
+import gymnasium as gym
+import numpy as np
+import torch
+import torch.nn as nn
+import torch.optim as optim
+import tyro
+from torch.distributions.categorical import Categorical
+
+import sys
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), '../../'))
+
+from ragen.env.rubikscube.env import RubiksCube2x2Env
+from ragen.env.rubikscube.config import RubiksCube2x2Config
+
+
+class RubiksCubeWrapper(gym.Env):
+    """
+    Adapter to use ragen RubiksCube2x2Env with Gymnasium vector API.
+    - Converts text observation to one-hot vector of 24 stickers x 6 colors.
+    - Maps agent actions [0..11] to env actions [1..12].
+    - Exposes proper observation_space and action_space.
+    """
+    metadata = {"render_modes": ["rgb_array", "human", "ansi"]}
+
+    def __init__(self, env: RubiksCube2x2Env):
+        super().__init__()
+        self._env = env
+        # 24 stickers, 6 colors -> one-hot size 144
+        self._colors = ['W', 'O', 'G', 'R', 'B', 'Y']
+        self._color_to_idx = {c: i for i, c in enumerate(self._colors)}
+        self.observation_space = gym.spaces.Box(low=0.0, high=1.0, shape=(24 * len(self._colors),), dtype=np.float32)
+        self.action_space = gym.spaces.Discrete(12)
+        # precompile regex to extract 4 letters within [X, X]\n [X, X]
+        # Lines look like: "Up (U):    [W, W]\n       [W, W]"
+        self._face_pat = re.compile(r"\[(?:\s*([A-Z])\s*,\s*([A-Z])\s*\])\n\s*\[(?:\s*([A-Z])\s*,\s*([A-Z])\s*\])")
+
+    def _encode_obs(self, text_obs: str) -> np.ndarray:
+        # Extract the six face blocks in the order U, L, F, R, B, D as rendered
+        faces_order = ["Up (U):", "Left (L):", "Front (F):", "Right (R):", "Back (B):", "Down (D):"]
+        onehots: List[int] = []
+        # Build a map from header to its block text
+        lines = text_obs.splitlines()
+        # Collect blocks starting after header line and including the next line for second row
+        i = 0
+        blocks: List[str] = []
+        while i < len(lines):
+            line = lines[i]
+            for header in faces_order:
+                if line.startswith(header):
+                    # Join current line's bracketed pair and the next line (which contains the second pair)
+                    # Remove the "Header:" prefix to keep only the bracket portions
+                    content = line[len(header):].strip()
+                    next_line = lines[i + 1] if i + 1 < len(lines) else ""
+                    block = f"{content}\n{next_line}"
+                    blocks.append(block)
+                    break
+            i += 1
+        # Fallback: if regex fails, return zeros
+        if len(blocks) != 6:
+            return np.zeros(24 * len(self._colors), dtype=np.float32)
+        stickers: List[int] = []
+        for blk in blocks:
+            m = self._face_pat.search(blk)
+            if not m:
+                return np.zeros(24 * len(self._colors), dtype=np.float32)
+            # order: 0,1,2,3 per render() doc
+            c0, c1, c2, c3 = m.group(1), m.group(2), m.group(3), m.group(4)
+            stickers.extend([c0, c1, c2, c3])
+        # stickers now length 24; convert to one-hot
+        grid = np.zeros((24, len(self._colors)), dtype=np.float32)
+        for idx, ch in enumerate(stickers):
+            cidx = self._color_to_idx.get(ch, None)
+            if cidx is not None:
+                grid[idx, cidx] = 1.0
+        return grid.reshape(-1)
+
+    def reset(self, *, seed: int | None = None, options: Dict[str, Any] | None = None):
+        text_obs = self._env.reset(seed=seed)
+        obs = self._encode_obs(text_obs)
+        return obs, {}
+
+    def step(self, action: int):
+        mapped = int(action) + 1  # 0..11 -> 1..12
+        text_obs, reward, done, info = self._env.step(mapped)
+        obs = self._encode_obs(text_obs)
+        terminated = bool(done)
+        truncated = False
+        return obs, float(reward), terminated, truncated, info or {}
+
+    def render(self):
+        return self._env.render()
+
+    def close(self):
+        self._env.close()
+
+
+@dataclass
+class Args:
+    exp_name: str = os.path.basename(__file__)[: -len(".py")]
+    seed: int = 1
+    torch_deterministic: bool = True
+    cuda: bool = True
+    track: bool = True
+    wandb_project_name: str = "cleanRL"
+    wandb_entity: str | None = None
+    capture_video: bool = False
+
+    # Algorithm
+    env_id: str = "RubiksCube2x2"
+    total_timesteps: int = 1000_000
+    learning_rate: float = 2.5e-4
+    num_envs: int = 8
+    num_steps: int = 128
+    anneal_lr: bool = True
+    gamma: float = 0.99
+    gae_lambda: float = 0.95
+    num_minibatches: int = 4
+    update_epochs: int = 4
+    norm_adv: bool = True
+    clip_coef: float = 0.2
+    clip_vloss: bool = True
+    ent_coef: float = 0.01
+    vf_coef: float = 0.5
+    max_grad_norm: float = 0.5
+    target_kl: float | None = None
+
+    # Rubik specific
+    scramble_depth: int = 2
+    max_steps_env: int = 20
+
+    # runtime filled
+    batch_size: int = 0
+    minibatch_size: int = 0
+    num_iterations: int = 0
+    eval_splits: int = 2
+    eval_episodes: int = 4000
+
+
+def make_env(idx, run_name, seed, scramble_depth, max_steps_env, capture_video=False):
+    def thunk():
+        config = RubiksCube2x2Config(scramble_depth=scramble_depth, max_steps=max_steps_env, render_mode='text')
+        env = RubiksCube2x2Env(config)
+        env = RubiksCubeWrapper(env)
+        env = gym.wrappers.TimeLimit(env, max_episode_steps=max_steps_env)
+        env = gym.wrappers.RecordEpisodeStatistics(env)
+        if capture_video and idx == 0:
+            env = gym.wrappers.RecordVideo(env, f"videos/{run_name}")
+        return env
+    return thunk
+
+
+def layer_init(layer, std=np.sqrt(2), bias_const=0.0):
+    torch.nn.init.orthogonal_(layer.weight, std)
+    torch.nn.init.constant_(layer.bias, bias_const)
+    return layer
+
+
+class Agent(nn.Module):
+    def __init__(self, envs):
+        super().__init__()
+        obs_shape = int(np.array(envs.single_observation_space.shape).prod())
+        hidden = 128
+        self.critic = nn.Sequential(
+            layer_init(nn.Linear(obs_shape, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, 1), std=1.0),
+        )
+        self.actor = nn.Sequential(
+            layer_init(nn.Linear(obs_shape, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, envs.single_action_space.n), std=0.01),
+        )
+
+    def get_value(self, x):
+        return self.critic(x)
+
+    def get_action_and_value(self, x, action=None):
+        logits = self.actor(x)
+        probs = Categorical(logits=logits)
+        if action is None:
+            action = probs.sample()
+        return action, probs.log_prob(action), probs.entropy(), self.critic(x)
+
+
+if __name__ == "__main__":
+    args = tyro.cli(Args)
+    args.batch_size = int(args.num_envs * args.num_steps)
+    args.minibatch_size = int(args.batch_size // args.num_minibatches)
+    args.num_iterations = args.total_timesteps // args.batch_size
+    run_name = f"{args.env_id}__{args.exp_name}__{args.seed}__{int(time.time())}"
+
+    if args.track:
+        import wandb
+        wandb.init(
+            project=args.wandb_project_name,
+            entity=args.wandb_entity,
+            config=vars(args),
+            name=run_name,
+            monitor_gym=True,
+            save_code=True,
+        )
+        try:
+            wandb.define_metric("global_step")
+            for prefix in ["train/*", "rollout/*", "eval/*", "losses/*", "charts/*", "perf/*"]:
+                wandb.define_metric(prefix, step_metric="global_step")
+        except Exception:
+            pass
+
+    # seeding
+    random.seed(args.seed)
+    np.random.seed(args.seed)
+    torch.manual_seed(args.seed)
+    torch.backends.cudnn.deterministic = args.torch_deterministic
+
+    device = torch.device("cuda" if torch.cuda.is_available() and args.cuda else "cpu")
+
+    # envs
+    envs = gym.vector.SyncVectorEnv([
+        make_env(i, run_name, args.seed, args.scramble_depth, args.max_steps_env, args.capture_video)
+        for i in range(args.num_envs)
+    ])
+    assert isinstance(envs.single_action_space, gym.spaces.Discrete)
+
+    agent = Agent(envs).to(device)
+    optimizer = optim.Adam(agent.parameters(), lr=args.learning_rate, eps=1e-5)
+
+    # storage
+    obs = torch.zeros((args.num_steps, args.num_envs) + envs.single_observation_space.shape).to(device)
+    actions = torch.zeros((args.num_steps, args.num_envs) + envs.single_action_space.shape).to(device)
+    logprobs = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    rewards = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    dones = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    values = torch.zeros((args.num_steps, args.num_envs)).to(device)
+
+    # start
+    global_step = 0
+    start_time = time.time()
+    next_obs, _ = envs.reset(seed=args.seed)
+    next_obs = torch.Tensor(next_obs).to(device)
+    next_done = torch.zeros(args.num_envs).to(device)
+
+    # eval helper similar to FrozenLake: collect greedy eval trajectories and write metrics.json
+    def collect_eval_trajectories(agent_model, make_env_fn, n_episodes, step_tag):
+        out_dir = Path(f"runs/{run_name}/trajectories/step_{step_tag}")
+        out_dir.mkdir(parents=True, exist_ok=True)
+        out_path = out_dir / "trajectories.jsonl"
+        env = make_env_fn()
+        collected = 0
+        summary_returns = []
+        summary_success = []
+        with out_path.open("w") as f:
+            while collected < n_episodes:
+                state, _ = env.reset(seed=args.seed + 100000 + collected)
+                traj_states = [state.tolist()]
+                traj_actions = []
+                traj_rewards = []
+                traj_dones = []
+                traj_success = []
+                done = False
+                step_count = 0
+                max_eval_steps = getattr(env, '_max_episode_steps', None) or int(args.max_steps_env)
+                while not done:
+                    with torch.no_grad():
+                        logits = agent_model.actor(torch.tensor(state, dtype=torch.float32, device=device).unsqueeze(0))
+                        action = int(torch.argmax(logits, dim=1).item())
+                    next_state, reward, terminated, truncated, info = env.step(action)
+                    traj_actions.append(int(action))
+                    traj_rewards.append(float(reward))
+                    step_count += 1
+                    d = bool(terminated) or bool(truncated) or (step_count >= max_eval_steps)
+                    traj_dones.append(d)
+                    traj_success.append(bool((info or {}).get('success', False)))
+                    state = next_state
+                    traj_states.append(state.tolist())
+                    done = d
+                ep_ret = float(sum(traj_rewards))
+                ep_succ = bool(any(traj_success))
+                record = {
+                    "states": traj_states,
+                    "actions": traj_actions,
+                    "rewards": traj_rewards,
+                    "dones": traj_dones,
+                    "success": traj_success,
+                    "episode_return": ep_ret,
+                    "episode_success": ep_succ,
+                }
+                f.write(json.dumps(record) + "\n")
+                collected += 1
+                summary_returns.append(ep_ret)
+                summary_success.append(1.0 if ep_succ else 0.0)
+        env.close()
+        try:
+            metrics = {
+                "global_step": int(step_tag),
+                "episodes": int(n_episodes),
+                "success_rate": float(np.mean(summary_success)) if len(summary_success) else 0.0,
+                "avg_return": float(np.mean(summary_returns)) if len(summary_returns) else 0.0,
+                "std_return": float(np.std(summary_returns)) if len(summary_returns) else 0.0,
+            }
+            with (out_dir / "metrics.json").open("w") as mf:
+                json.dump(metrics, mf)
+        except Exception as e:
+            print(f"Warning: failed to write eval metrics: {e}")
+
+    # training loop
+    eval_every_iters = max(1, args.num_iterations // args.eval_splits)
+    for iteration in range(1, args.num_iterations + 1):
+        # Anneal LR
+        if args.anneal_lr:
+            frac = 1.0 - (iteration - 1.0) / args.num_iterations
+            lrnow = frac * args.learning_rate
+            optimizer.param_groups[0]["lr"] = lrnow
+
+        # accumulate per-iteration episode successes
+        iter_successes = []
+        for step in range(0, args.num_steps):
+            global_step += args.num_envs
+            obs[step] = next_obs
+            dones[step] = next_done
+
+            with torch.no_grad():
+                action, logprob, _, value = agent.get_action_and_value(next_obs)
+                values[step] = value.flatten()
+            actions[step] = action
+            logprobs[step] = logprob
+
+            next_obs, reward, terminations, truncations, infos = envs.step(action.cpu().numpy())
+            next_done = np.logical_or(terminations, truncations)
+            rewards[step] = torch.tensor(reward).to(device).view(-1)
+            next_obs, next_done = torch.Tensor(next_obs).to(device), torch.Tensor(next_done).to(device)
+
+            # Episode stats logging similar to FrozenLake
+            try:
+                mask = None
+                if isinstance(infos, dict):
+                    if "_episode" in infos:
+                        mask = np.asarray(infos["_episode"]).astype(bool)
+                    elif "episode" in infos and isinstance(infos["episode"], dict) and "_l" in infos["episode"]:
+                        mask = np.asarray(infos["episode"]["_l"]).astype(bool)
+                if mask is not None and np.any(mask):
+                    r_arr = np.asarray(infos.get("episode", {}).get("r", np.zeros_like(mask, dtype=float)))
+                    l_arr = np.asarray(infos.get("episode", {}).get("l", np.zeros_like(mask, dtype=int)))
+                    # prefer success from info; fallback to ep return > 0
+                    if "success" in infos:
+                        succ_arr = np.asarray(infos.get("success", np.zeros_like(mask, dtype=bool))).astype(float)
+                    else:
+                        try:
+                            succ_arr = (np.asarray(r_arr) > 0).astype(float)
+                        except Exception:
+                            succ_arr = np.zeros_like(mask, dtype=float)
+                    # collect iteration successes for training success rate
+                    try:
+                        for s in np.asarray(succ_arr)[mask]:
+                            iter_successes.append(float(s))
+                    except Exception:
+                        pass
+                    if args.track:
+                        try:
+                            import wandb
+                            log_dict = {
+                                "global_step": int(global_step),
+                                "rollout/ep_rew_mean": float(np.mean(r_arr[mask])) if np.any(mask) else None,
+                                "rollout/ep_len_mean": float(np.mean(l_arr[mask])) if np.any(mask) else None,
+                                "rollout/success_rate": float(np.mean(succ_arr[mask])) if np.any(mask) else None,
+                            }
+                            wandb.log(log_dict, step=global_step)
+                        except Exception:
+                            pass
+            except Exception:
+                pass
+
+        # GAE
+        with torch.no_grad():
+            next_value = agent.get_value(next_obs).reshape(1, -1)
+            advantages = torch.zeros_like(rewards).to(device)
+            lastgaelam = 0
+            for t in reversed(range(args.num_steps)):
+                if t == args.num_steps - 1:
+                    nextnonterminal = 1.0 - next_done
+                    nextvalues = next_value
+                else:
+                    nextnonterminal = 1.0 - dones[t + 1]
+                    nextvalues = values[t + 1]
+                delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
+                advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
+            returns = advantages + values
+
+        # flatten batch
+        b_obs = obs.reshape((-1,) + envs.single_observation_space.shape)
+        b_logprobs = logprobs.reshape(-1)
+        b_actions = actions.reshape((-1,) + envs.single_action_space.shape)
+        b_advantages = advantages.reshape(-1)
+        b_returns = returns.reshape(-1)
+        b_values = values.reshape(-1)
+
+        # update
+        b_inds = np.arange(args.batch_size)
+        clipfracs = []
+        for epoch in range(args.update_epochs):
+            np.random.shuffle(b_inds)
+            for start in range(0, args.batch_size, args.minibatch_size):
+                end = start + args.minibatch_size
+                mb_inds = b_inds[start:end]
+
+                _, newlogprob, entropy, newvalue = agent.get_action_and_value(b_obs[mb_inds], b_actions.long()[mb_inds])
+                logratio = newlogprob - b_logprobs[mb_inds]
+                ratio = logratio.exp()
+
+                with torch.no_grad():
+                    old_approx_kl = (-logratio).mean()
+                    approx_kl = ((ratio - 1) - logratio).mean()
+                    clipfracs += [((ratio - 1.0).abs() > args.clip_coef).float().mean().item()]
+
+                mb_advantages = b_advantages[mb_inds]
+                if args.norm_adv:
+                    mb_advantages = (mb_advantages - mb_advantages.mean()) / (mb_advantages.std() + 1e-8)
+
+                pg_loss1 = -mb_advantages * ratio
+                pg_loss2 = -mb_advantages * torch.clamp(ratio, 1 - args.clip_coef, 1 + args.clip_coef)
+                pg_loss = torch.max(pg_loss1, pg_loss2).mean()
+
+                newvalue = newvalue.view(-1)
+                if args.clip_vloss:
+                    v_loss_unclipped = (newvalue - b_returns[mb_inds]) ** 2
+                    v_clipped = b_values[mb_inds] + torch.clamp(
+                        newvalue - b_values[mb_inds], -args.clip_coef, args.clip_coef,
+                    )
+                    v_loss_clipped = (v_clipped - b_returns[mb_inds]) ** 2
+                    v_loss = 0.5 * torch.max(v_loss_unclipped, v_loss_clipped).mean()
+                else:
+                    v_loss = 0.5 * ((newvalue - b_returns[mb_inds]) ** 2).mean()
+
+                entropy_loss = entropy.mean()
+                loss = pg_loss - args.ent_coef * entropy_loss + v_loss * args.vf_coef
+
+                optimizer.zero_grad()
+                loss.backward()
+                nn.utils.clip_grad_norm_(agent.parameters(), args.max_grad_norm)
+                optimizer.step()
+
+            if args.target_kl is not None and approx_kl > args.target_kl:
+                break
+
+        # metrics
+        y_pred, y_true = b_values.cpu().numpy(), b_returns.cpu().numpy()
+        var_y = np.var(y_true)
+        explained_var = np.nan if var_y == 0 else 1 - np.var(y_true - y_pred) / var_y
+
+        sps = int(global_step / (time.time() - start_time))
+        train_success_rate = float(np.mean(iter_successes)) if len(iter_successes) else 0.0
+        print(f"Iter {iteration:4d}/{args.num_iterations} | SPS: {sps:5d} | R: {rewards.mean().item():6.3f}")
+        if args.track:
+            try:
+                import wandb
+                wandb.log({
+                    "global_step": int(global_step),
+                    "charts/progress": float(100.0 * iteration / max(1, args.num_iterations)),
+                    "train/value_loss": float(v_loss.item()),
+                    "train/policy_loss": float(pg_loss.item()),
+                    "losses/value_loss": float(v_loss.item()),
+                    "losses/policy_loss": float(pg_loss.item()),
+                    "train/entropy": float(entropy_loss.item()),
+                    "train/old_approx_kl": float(old_approx_kl.item()),
+                    "train/approx_kl": float(approx_kl.item()),
+                    "train/clipfrac": float(np.mean(clipfracs)) if len(clipfracs) else 0.0,
+                    "losses/explained_variance": float(explained_var),
+                    "charts/avg_reward": float(rewards.mean().item()),
+                    "charts/avg_value": float(values.mean().item()),
+                    "perf/SPS": int(sps),
+                    "charts/SPS": int(sps),
+                    "train/success_rate": train_success_rate,
+                    "charts/train_success_rate": train_success_rate,
+                    "train/learning_rate": float(optimizer.param_groups[0]["lr"]),
+                }, step=global_step)
+            except Exception:
+                pass
+
+        # periodic evaluation collection
+        if iteration % eval_every_iters == 0:
+            try:
+                eval_thunk = make_env(0, run_name, args.seed + 9999, args.scramble_depth, args.max_steps_env, False)
+                collect_eval_trajectories(agent, eval_thunk, n_episodes=args.eval_episodes, step_tag=global_step)
+                if args.track:
+                    try:
+                        import json as _json
+                        from pathlib import Path as _Path
+                        mpath = _Path(f"runs/{run_name}/trajectories/step_{global_step}/metrics.json")
+                        if mpath.exists():
+                            with mpath.open("r") as mf:
+                                metrics = _json.load(mf)
+                            wandb.log({
+                                "eval/success_rate": metrics.get("success_rate"),
+                                "eval/avg_return": metrics.get("avg_return"),
+                                "eval/std_return": metrics.get("std_return"),
+                                "eval/episodes": metrics.get("episodes"),
+                            }, step=global_step)
+                    except Exception:
+                        pass
+                print(f"Collected {args.eval_episodes} eval trajectories at global_step {global_step}")
+            except Exception as e:
+                print(f"Warning: eval trajectory collection failed at step {global_step}: {e}")
+
+    envs.close()

wandb/run-20251218_184253-g8kppnjh/files/config.yaml

@@ -0,0 +1,201 @@
+_wandb:
+    value:
+        cli_version: 0.23.0
+        code_path: code/cleanrl/cleanrl/ppo_rubikscube.py
+        e:
+            xtmqsnhbpijxn98k1keqte2hruuy3myj:
+                args:
+                    - --track
+                codePath: cleanrl/cleanrl/ppo_rubikscube.py
+                codePathLocal: cleanrl/cleanrl/ppo_rubikscube.py
+                cpu_count: 64
+                cpu_count_logical: 128
+                cudaVersion: "12.4"
+                disk:
+                    /:
+                        total: "5153960755200"
+                        used: "447529222144"
+                email: haoyu-wa22@mails.tsinghua.edu.cn
+                executable: /root/local/miniconda3/envs/ragen/bin/python
+                git:
+                    commit: 8d73639b99b38265453f898b8d6af7d4af50d56e
+                    remote: https://github.com/mll-lab-nu/RAGEN.git
+                gpu: NVIDIA H100 80GB HBM3
+                gpu_count: 8
+                gpu_nvidia:
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-35e2d43d-4067-82ce-90d4-def9e389bf28
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-af4135e3-88f2-e9ac-518d-502c75a85429
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-d7fdeeba-fe9b-ec03-d9f7-6724fe4266b5
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-ccc4f668-3882-5a8e-2c07-c5cd08f6f666
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-7b73c0cf-d3d5-e10c-7176-a43be1e41001
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-81b58d94-5d1f-8ec2-f9d2-fd56172ed177
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-03e8bc66-3b44-6794-49fd-5392fbdda6d1
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-b7cf0ec6-7c29-1179-dceb-09565da51890
+                host: pt-4d654cf4576f4d23ad3d3919f12932fe-worker-0
+                memory:
+                    total: "2163642122240"
+                os: Linux-5.14.0-284.25.1.el9_2.x86_64-x86_64-with-glibc2.35
+                program: /mnt/general/wanghy/RAGEN/cleanrl/cleanrl/ppo_rubikscube.py
+                python: CPython 3.10.19
+                root: /mnt/general/wanghy/RAGEN
+                startedAt: "2025-12-18T10:42:53.636495Z"
+                writerId: xtmqsnhbpijxn98k1keqte2hruuy3myj
+        m:
+            - "1": global_step
+              "6":
+                - 3
+              "7": []
+            - "2": train/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": rollout/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": eval/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": losses/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": charts/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": perf/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+        python_version: 3.10.19
+        t:
+            "1":
+                - 1
+                - 11
+                - 30
+                - 49
+                - 50
+                - 51
+                - 105
+            "2":
+                - 1
+                - 11
+                - 30
+                - 49
+                - 50
+                - 51
+                - 105
+            "3":
+                - 7
+                - 13
+                - 16
+                - 61
+            "4": 3.10.19
+            "5": 0.23.0
+            "6": 4.57.1
+            "12": 0.23.0
+            "13": linux-x86_64
+anneal_lr:
+    value: true
+batch_size:
+    value: 1024
+capture_video:
+    value: false
+clip_coef:
+    value: 0.2
+clip_vloss:
+    value: true
+cuda:
+    value: true
+ent_coef:
+    value: 0.01
+env_id:
+    value: RubiksCube2x2
+eval_episodes:
+    value: 4000
+eval_splits:
+    value: 2
+exp_name:
+    value: ppo_rubikscube
+gae_lambda:
+    value: 0.95
+gamma:
+    value: 0.99
+learning_rate:
+    value: 0.00025
+max_grad_norm:
+    value: 0.5
+max_steps_env:
+    value: 20
+minibatch_size:
+    value: 256
+norm_adv:
+    value: true
+num_envs:
+    value: 8
+num_iterations:
+    value: 976
+num_minibatches:
+    value: 4
+num_steps:
+    value: 128
+scramble_depth:
+    value: 2
+seed:
+    value: 1
+target_kl:
+    value: null
+torch_deterministic:
+    value: true
+total_timesteps:
+    value: 1000000
+track:
+    value: true
+update_epochs:
+    value: 4
+vf_coef:
+    value: 0.5
+wandb_entity:
+    value: null
+wandb_project_name:
+    value: cleanRL

wandb/run-20251218_184253-g8kppnjh/files/diff.patch

@@ -0,0 +1,1162 @@
+diff --git a/config/base.yaml b/config/base.yaml
+index a1659e4..9df6250 100644
+--- a/config/base.yaml
++++ b/config/base.yaml
+@@ -9,15 +9,16 @@ seed:
+   train: 10000
+   val: 123
+ 
+-micro_batch_size_per_gpu: 4
++micro_batch_size_per_gpu: 1
+ ppo_mini_batch_size: 32
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/Qwen/Qwen2.5-0.5B-Instruct
++# Qwen/Qwen2.5-0.5B-Instruct
+ enable_response_mask: True # Enabling response mask could improve stability of rollout/old_log_prob, as P(st|history) are no longer calculated in loss here. See https://docs.google.com/document/d/1bg7obeiKTExuHHBl5uOiSpec5uLDZ2Tgvxy6li5pHX4/edit?usp=sharing for more details.
+ grpo_advantage_length_weight: False # if you do not enable this and critic/advantage_estimator is GRPO, and the critic/advantages/mean is too low, then you can try enabling this to encourage reasoning and forbid collapse
+ 
+ lora:
+   rank: 0
+-  alpha: 64
++  alpha: 16
+   target_modules: all-linear
+ 
+ actor_rollout_ref:
+@@ -46,10 +47,10 @@ actor_rollout_ref:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu} # following micro_batch_size_per_gpu
+     tensor_model_parallel_size: 1
+-    max_model_len: 3600
++    max_model_len: 7200 #3600 why** 14400
+     prompt_length: 1 # useless. Just put it here
+     response_length: 400 # single-turn response length
+-    gpu_memory_utilization: 0.5
++    gpu_memory_utilization: 0.7
+     max_num_batched_tokens: 8192 # set only when enable_chunked_prefill is true
+     temperature: 1
+     rollout_filter_ratio: 0.25
+@@ -90,26 +91,28 @@ algorithm:
+     kl_coef: 0.000
+ 
+ trainer:
+-  project_name: ragen_latest
++  project_name: 
+   experiment_name: test
+   local_log_dir: "results/"
+-  save_freq: 100
++  save_freq: -1
+   total_training_steps: 200
+   validation_steps: 1 # validation instances = validation_steps * val_env_groups * group_size
+   val_before_train: True
+   n_gpus_per_node: 1
+   test_freq: 10
+   generations_to_log_to_wandb: 
++    train: 128
+     val: 20
+   logger: [ 'console', 'wandb' ]
+   max_actor_ckpt_to_keep: 1
+   max_critic_ckpt_to_keep: 1
++  default_local_dir: /mnt/general/wanghy/RAGEN/saves/
+ 
+ agent_proxy:
+   max_context_window: -1 # set a value > 0 to enable context window for long trajectory
+-  max_turn: 5
++  max_turn: 25 #25 why** 700
+   action_sep: "||"
+-  max_actions_per_turn: 2 # how many actions can be output at most in a single turn
++  max_actions_per_turn: 1 # how many actions can be output at most in a single turn
+   use_turn_scores: False # important to GAE when applying token-level rewards to token-level advantages. If False, will take the sum of scores as the reward for the last turn.
+   enable_think: True # False -> no think RL
+   reward_normalization:
+@@ -126,11 +129,11 @@ es_manager:
+       tags: ["CoordSokoban"]
+       n_groups: [8] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+   val:
+-    env_groups: 32
+-    group_size: 16 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
++    env_groups: 256
++    group_size: 1 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
+     env_configs:
+       tags: ["CoordSokoban"]
+-      n_groups: [32] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      n_groups: [256] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ ctx_manager:
+   generation: # go to vllm
+diff --git a/config/envs.yaml b/config/envs.yaml
+index 1002ac8..ba3ac6f 100644
+--- a/config/envs.yaml
++++ b/config/envs.yaml
+@@ -2,6 +2,7 @@ custom_envs:
+   SimpleSokoban:
+     env_type: sokoban
+     max_actions_per_traj: 10 # used in environment state manager to control the actual max actions executed per trajectory
++    # ORIGNAL env_instruction
+     env_instruction: |
+       You are solving the Sokoban puzzle. 
+       You are the player and you need to push all boxes to targets. 
+@@ -12,7 +13,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+ 
+   LargerSokoban:
+@@ -65,7 +66,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+       observation_format: "grid_coord"
+ 
+@@ -111,14 +112,14 @@ custom_envs:
+ 
+   FrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: "You are solving the FrozenLake puzzle. Forbid the whole and go to the target. You may move to the unintended direction due to the slippery ice. Example answer format: <think>To forbid the hole and go to the target, I should go left then go up.</think><answer>Left || Up</answer>"
+     max_tokens: 100
+     env_config: null
+ 
+   CoordFrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: |
+       You are solving the FrozenLake puzzle. The observation includes both a symbol grid and zero-indexed coordinates for the start, goal, player, and any holes.
+       Coordinates range from the top-left corner (0, 0) to the bottom-right corner (5, 5).
+@@ -195,4 +196,80 @@ custom_envs:
+     max_actions_per_traj: 30
+     env_instruction: "You are a Lean theorem prover. Given a Lean theorem statement, propose a sequence of tactics that completes the proof. Think step by step about which tactics to apply next. Provide tactics separated by '||', for example <answer>intro || simp || rfl</answer>."
+     max_tokens: 512
+-    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
+
++    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
++
++  
++  game_2048:
++    env_type: game_2048
++    max_actions_per_traj: 700
++    env_instruction: |
++      You are playing the 2048 game on a 4x4 grid. Merge equal tiles by sliding Up, Right, Down, or Left.
++      If a move is invalid (no tiles move), a small penalty is applied. Respond with a single action.
++      Example: <answer>Up</answer>
++    max_tokens: 8192
++    env_config: null
++
++  blackjack:
++    env_type: blackjack
++    max_actions_per_traj: 10
++    env_instruction: |
++      You are playing Blackjack against a dealer. The dealer must hit on 16 or less and stand on 17 or more.
++      Choose either Stick or Hit. Respond with a single action.
++      Example: <answer>Hit</answer>
++    max_tokens: 64
++    env_config: null
++
++  rubikscube:
++    env_type: rubikscube
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.
++      Available actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.
++      - Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).
++      - Modifiers: A letter alone means 90° clockwise (e.g., 'R'). A letter with prime (') means 90° counter-clockwise (e.g., "R'").
++      Respond with a sequence of actions separated by "||".
++      Example: <answer>U</answer>
++    max_tokens: 96
++    env_config:
++      scramble_depth: 1
++      max_steps: 20
++      render_mode: "text"
++  
++  SimpleSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 2x2 box contains the numbers 1-4 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format, for example: <answer>place 1 at row 2 col 3</answer> or <answer>1,2,3</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 4
++      difficulty: "easy"
++      render_format: "with_feedback"
++      show_conflicts: false
++      show_valid_numbers: false
++      max_steps: 20
++
++  MediumSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 30
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 3x3 box contains the numbers 1-9 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format: <answer>place 5 at row 2 col 3</answer> or <answer>2,3,5</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 9
++      difficulty: "medium"
++      render_format: "with_feedback"
++      show_conflicts: true
++      show_valid_numbers: true
++      max_steps: 81
++
+diff --git a/config/eval.yaml b/config/eval.yaml
+index e65f3a5..7775fd1 100644
+--- a/config/eval.yaml
++++ b/config/eval.yaml
+@@ -1,14 +1,15 @@
+ defaults:
+   - envs
+-
++enable_response_mask: True
+ system:
+-  CUDA_VISIBLE_DEVICES: "0"
++  CUDA_VISIBLE_DEVICES: "0,1,2,3,4,5,6,7"
+ 
+ seed:
+   train: 10000
+   val: 123
+ 
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/openai/gpt-oss-20b
++# /mnt/general/share/model/tyzhu/SPA-frozenlake-qwen2.5-1.5b-instruct
+ 
+ lora:
+   rank: 0
+@@ -24,9 +25,9 @@ actor_rollout_ref:
+   rollout:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu}
+-    tensor_model_parallel_size: 1
++    tensor_model_parallel_size: 8
+     dtype: bfloat16
+-    max_model_len: 3600
++    max_model_len: 7200
+     prompt_length: 1
+     response_length: 400
+     gpu_memory_utilization: 0.9
+@@ -35,6 +36,7 @@ actor_rollout_ref:
+     free_cache_engine: True
+     enable_chunked_prefill: False
+     disable_log_stats: False
++    do_sample: True
+     val_kwargs:
+       do_sample: True
+       temperature: 0.5
+@@ -44,9 +46,9 @@ actor_rollout_ref:
+ 
+ agent_proxy:
+   max_context_window: -1
+-  max_turn: 5
++  max_turn: 25
+   action_sep: "||"
+-  max_actions_per_turn: 2
++  max_actions_per_turn: 1
+   use_turn_scores: False
+   enable_think: True
+   reward_normalization:
+@@ -59,13 +61,13 @@ es_manager:
+     env_groups: 8
+     group_size: 16
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [8]
+   val:
+     env_groups: 32
+-    group_size: 16
++    group_size: 128
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [32]
+ 
+ ctx_manager:
+diff --git a/config/evaluate_api_llm.yaml b/config/evaluate_api_llm.yaml
+index b21c95e..e2e6761 100644
+--- a/config/evaluate_api_llm.yaml
++++ b/config/evaluate_api_llm.yaml
+@@ -1,8 +1,11 @@
++#export OPENAI_BASE_URL="https://api.ohmygpt.com/v1"
++#export OPENAI_API_KEY="sk-o4sMxBkN5BB100C4D4a3T3BlBkFJF7791CA39EA14ca98041"
++#python -m ragen.eval_api hydra.searchpath='[file://./verl/verl/trainer/config]'
+ defaults:
+   - base # this is a symbolic link to the verl/verl/trainer/config/ppo_trainer.yaml file
+ 
+ model_config:
+-  model_name: gpt-4o # should be registered in model_info
++  model_name: ark-deepseek-v3-250324 # should be registered in model_info
+   max_concurrency: 16
+ 
+ model_info:
+@@ -24,9 +27,9 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+-  gpt-4o:
++  gpt-4o-mini:
+     provider_name: openai
+-    model_name: gpt-4o
++    model_name: gpt-4o-mini
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+@@ -36,21 +39,42 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  ark-deepseek-v3-250324:
++    provider_name: openai
++    model_name: ark-deepseek-v3-250324
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
+   deepseek-v3:
+     provider_name: deepseek
+     model_name: deepseek-chat
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  glm-4.6:
++    provider_name: openai
++    model_name: glm-4.6
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
++  TA/openai/gpt-oss-120b:
++    provider_name: openai
++    model_name: TA/openai/gpt-oss-120b
++    generation_kwargs:
++      temperature: 0
++      max_tokens: 8192
++      # max_retries: 5
+ 
+-
+-
++agent_proxy:
++  max_turn: 5
+ es_manager:
+   val:
+-    env_groups: 256
++    env_groups: 128
+     group_size: 1 # should be set to 1 because val temperature is set to 0 and same prompt leads to same output
+     env_configs:
+-      tags: ["CoordSokoban"]
+-      n_groups: [256] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      tags: ["rubikscube"]
++      n_groups: [128] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ 
++rollout:
++  max_model_len: 7200
+
+Submodule external/webshop-minimal contains modified content
+diff --git a/external/webshop-minimal/requirements.txt b/external/webshop-minimal/requirements.txt
+index 5a1b04f..238ed5a 100644
+--- a/external/webshop-minimal/requirements.txt
++++ b/external/webshop-minimal/requirements.txt
+@@ -4,7 +4,7 @@ flask
+ html2text
+ rank_bm25
+ pyserini
+-faiss-cpu
++faiss-gpu
+ thefuzz
+ gdown
+ spacy
+diff --git a/ragen/env/__init__.py b/ragen/env/__init__.py
+index b0f3461..bc6838e 100644
+--- a/ragen/env/__init__.py
++++ b/ragen/env/__init__.py
+@@ -12,6 +12,14 @@ from .metamathqa.env import MetaMathQAEnv
+ from .metamathqa.config import MetaMathQAEnvConfig
+ from .lean.config import LeanEnvConfig
+ from .lean.env import LeanEnv
++from .game_2048.config import Game2048EnvConfig
++from .game_2048.env import Game2048Env
++from .blackjack.config import BlackjackEnvConfig
++from .blackjack.env import BlackjackEnv
++from .rubikscube.config import RubiksCube2x2Config
++from .rubikscube.env import RubiksCube2x2Env
++from .sudoku.config import SudokuEnvConfig
++from .sudoku.env import SudokuEnv
+ 
+ 
+ REGISTERED_ENVS = {
+@@ -22,6 +30,10 @@ REGISTERED_ENVS = {
+     # 'alfworld': AlfredTXTEnv,
+     'metamathqa': MetaMathQAEnv,
+     'lean': LeanEnv,
++    'game_2048': Game2048Env,
++    'blackjack': BlackjackEnv,
++    'rubikscube': RubiksCube2x2Env,
++    'sudoku': SudokuEnv,
+ }
+ 
+ REGISTERED_ENV_CONFIGS = {
+@@ -32,6 +44,10 @@ REGISTERED_ENV_CONFIGS = {
+     # 'alfworld': AlfredEnvConfig,
+     'metamathqa': MetaMathQAEnvConfig,
+     'lean': LeanEnvConfig,
++    'game_2048': Game2048EnvConfig,
++    'blackjack': BlackjackEnvConfig,
++    'rubikscube': RubiksCube2x2Config,
++    'sudoku': SudokuEnvConfig,
+ }
+ 
+ try:
+diff --git a/ragen/env/frozen_lake/config.py b/ragen/env/frozen_lake/config.py
+index de054f4..9950c34 100644
+--- a/ragen/env/frozen_lake/config.py
++++ b/ragen/env/frozen_lake/config.py
+@@ -8,7 +8,7 @@ class FrozenLakeEnvConfig:
+     size: int = 4
+     p: float = 0.9
+     success_rate: float = 0.8
+-    is_slippery: bool = True
++    is_slippery: bool = False
+     map_seed: Optional[int] = None
+     render_mode: str = "text"
+     observation_format: str = "grid"
+diff --git a/ragen/env/frozen_lake/env.py b/ragen/env/frozen_lake/env.py
+index 9ee3add..eef11d8 100644
+--- a/ragen/env/frozen_lake/env.py
++++ b/ragen/env/frozen_lake/env.py
+@@ -13,6 +13,7 @@ from ragen.env.base import BaseDiscreteActionEnv
+ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+     def __init__(self, config: FrozenLakeEnvConfig = FrozenLakeEnvConfig()):
+         # Using mappings directly from config
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.GRID_LOOKUP = config.grid_lookup
+         self.ACTION_LOOKUP = config.action_lookup
+@@ -95,6 +96,7 @@ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+ if __name__ == "__main__":
+     import matplotlib.pyplot as plt
+     config = FrozenLakeEnvConfig(size=4, is_slippery=True)
++    # import pdb;pdb.set_trace()
+     env = FrozenLakeEnv(config)
+     print(env.reset())
+     while True:
+diff --git a/ragen/env/sokoban/env.py b/ragen/env/sokoban/env.py
+index 17cd636..6922b54 100644
+--- a/ragen/env/sokoban/env.py
++++ b/ragen/env/sokoban/env.py
+@@ -33,6 +33,7 @@ class SokobanEnv(BaseDiscreteActionEnv, GymSokobanEnv):
+     def reset(self, seed=None, mode=None):
+         try:
+             with all_seed(seed):
++                # import pdb;pdb.set_trace()
+                 self.room_fixed, self.room_state, self.box_mapping, action_sequence = generate_room(
+                     dim=self.dim_room,
+                     num_steps=self.num_gen_steps,
+diff --git a/ragen/llm_agent/agent_proxy.py b/ragen/llm_agent/agent_proxy.py
+index c25b50d..f6c4088 100644
+--- a/ragen/llm_agent/agent_proxy.py
++++ b/ragen/llm_agent/agent_proxy.py
+@@ -287,6 +287,7 @@ def main(config):
+ 	proxy = LLMAgentProxy(config, actor_wg, tokenizer)
+ 	import time
+ 	start_time = time.time()
++	# import pdb;pdb.set_trace()
+ 	rollouts = proxy.rollout(
+ 		DataProto(
+ 			batch=None,
+diff --git a/ragen/llm_agent/base_llm.py b/ragen/llm_agent/base_llm.py
+index 358a2eb..b6f7797 100644
+--- a/ragen/llm_agent/base_llm.py
++++ b/ragen/llm_agent/base_llm.py
+@@ -33,6 +33,7 @@ class OpenAIProvider(LLMProvider):
+             raise ValueError("OpenAI API key not provided and not found in environment variables")
+         
+         self.client = AsyncOpenAI(api_key=self.api_key)
++        # import pdb;pdb.set_trace()
+     
+     async def generate(self, messages: List[Dict[str, str]], **kwargs) -> LLMResponse:
+         if "o1-mini" in self.model_name:
+@@ -46,10 +47,7 @@ class OpenAIProvider(LLMProvider):
+         )
+         if response.choices[0].finish_reason in ['length', 'content_filter']:
+             raise ValueError("Content filtered or length exceeded")
+-        return LLMResponse(
+-            content=response.choices[0].message.content,
+-            model_name=response.model
+-        )
++        return LLMResponse(content=response.choices[0].message.content,model_name=response.model)
+ 
+ class DeepSeekProvider(LLMProvider):
+     """DeepSeek API provider implementation"""
+@@ -199,7 +197,7 @@ class ConcurrentLLM:
+         
+         # Queue to store unfinished or failed tasks
+         current_batch = messages_list.copy()
+-        max_retries = kwargs.get("max_retries", 100)
++        max_retries = kwargs.get("max_retries", 10)
+         retry_count = 0
+         
+         while current_batch and retry_count < max_retries:
+diff --git a/ragen/llm_agent/ctx_manager.py b/ragen/llm_agent/ctx_manager.py
+index 905247a..af20a01 100644
+--- a/ragen/llm_agent/ctx_manager.py
++++ b/ragen/llm_agent/ctx_manager.py
+@@ -88,6 +88,7 @@ class ContextManager:
+         Initialize the ContextManager.
+         Processor is used to process the image data.
+         """
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.tokenizer = tokenizer
+         self.processor = processor
+@@ -311,7 +312,7 @@ class ContextManager:
+ 
+             llm_input_texts.append(text_with_prompt)
+             messages_list.append(messages)
+-
++        # import pdb;pdb.set_trace()
+         inputs = self.tokenizer(llm_input_texts, return_tensors="pt", padding=True, padding_side="left", truncation=False) # We have truncated previously, truncation in tokenizer may cause issues.
+         input_ids, attention_mask = inputs.input_ids, inputs.attention_mask
+         position_ids = (attention_mask.cumsum(dim=-1) - 1).clamp(min=0)
+@@ -353,6 +354,30 @@ class ContextManager:
+                 key: np.sum(value) / self.env_nums[key.split("/")[0]]
+                 for key, value in metrics.items()
+             }
++            # Derived metrics for wandb logging
++            try:
++                # charts/avg_episode_return: average across all 2048 env groups
++                ep_keys = [k for k in metrics.keys() if k.endswith('/episodic_return')]
++                if len(ep_keys) > 0:
++                    ep_vals = []
++                    for k in ep_keys:
++                        tag = k.split('/')[0]
++                        denom = self.env_nums.get(tag, max(1, len(metrics[k])))
++                        ep_vals.append(float(np.sum(metrics[k]) / denom))
++                    mean_metrics["charts/avg_episode_return"] = float(np.mean(ep_vals))
++            except Exception:
++                pass
++            try:
++                # rollout/max_tile: max over all envs in this batch
++                tile_keys = [k for k in metrics.keys() if k.endswith('/max_tile')]
++                if len(tile_keys) > 0:
++                    tile_vals = []
++                    for k in tile_keys:
++                        tile_vals.extend(metrics[k])
++                    if len(tile_vals) > 0:
++                        mean_metrics["rollout/max_tile"] = int(np.max(tile_vals))
++            except Exception:
++                pass
+             for key, values in metrics.items():
+                 if not isinstance(values, list):
+                     continue
+diff --git a/ragen/llm_agent/es_manager.py b/ragen/llm_agent/es_manager.py
+index b87a2b3..1a49d3e 100644
+--- a/ragen/llm_agent/es_manager.py
++++ b/ragen/llm_agent/es_manager.py
+@@ -128,18 +128,40 @@ class EnvStateManager:
+         env_outputs: List[Dict]
+             {env_id: int, history: List[Dict][{state: str, actions: List[str], reward: float, info: Dict, llm_response: str, llm_raw_response: str, (Optional)images: List[PIL.Image.Image]}]}
+         """
++        # def _execute_actions(env, actions):
++        #     acc_reward, turn_info, turn_done = 0, {}, False
++        #     executed_actions = []
++        #     for action in actions:
++        #         _, reward, done, info = env.step(action)
++        #         acc_reward += reward
++        #         turn_info.update(info) # NOTE: currently use last info for multi-action
++        #         executed_actions.append(action)
++        #         if done:
++        #             turn_done = True
++        #             break
++            
++        #     return acc_reward, turn_info, turn_done, executed_actions
+         def _execute_actions(env, actions):
+-            acc_reward, turn_info, turn_done = 0, {}, False
++            acc_reward, turn_info, turn_done = 0.0, {}, False
++            raw_acc_reward = 0.0
+             executed_actions = []
+             for action in actions:
+                 _, reward, done, info = env.step(action)
+-                acc_reward += reward
++                acc_reward += float(reward)
++                try:
++                    raw_acc_reward += float(info.get('raw_reward', 0.0))
++                except Exception:
++                    pass
+                 turn_info.update(info) # NOTE: currently use last info for multi-action
+                 executed_actions.append(action)
+                 if done:
+                     turn_done = True
+                     break
+-            
++            # Overwrite per-turn raw_reward to reflect the sum across all executed actions in this turn
++            try:
++                turn_info['raw_reward'] = float(raw_acc_reward)
++            except Exception:
++                pass
+             return acc_reward, turn_info, turn_done, executed_actions
+ 
+         def _log_env_state(status, history, cur_obs, max_actions_per_traj, executed_actions, all_actions, acc_reward, turn_done, turn_info, env_input):
+@@ -198,6 +220,18 @@ class EnvStateManager:
+                 'success': float(status.terminated and (not status.truncated)),
+                 'num_actions': status.num_actions,
+             }
++            # Add episodic-level metrics
++            # try:
++            #     # Sum of per-turn rewards equals the episodic return (env-shaped reward)
++            #     env_metric['episodic_return'] = float(sum(status.rewards))
++            # except Exception:
++            #     pass
++            try:
++                # Final max tile on the board at the end of the rollout
++                import numpy as _np
++                env_metric['max_tile'] = int(_np.max(entry['env'].grid))
++            except Exception:
++                pass
+             custom_metric = {}
+             for turn in cache['history']:
+                 for k, v in turn.get('info', {}).items():
+@@ -212,6 +246,12 @@ class EnvStateManager:
+                             "Skipping non-numeric metric '%s' with value %r for env %s.",
+                             k, v, entry['tag']
+                         )
++            # Add episodic_return as the SUM of raw_reward across steps (align with CleanRL)
++            try:
++                if 'raw_reward' in custom_metric:
++                    env_metric['episodic_return'] = float(np.sum(custom_metric['raw_reward']))
++            except Exception:
++                pass
+             for k, v in custom_metric.items():
+                 # TODO: Move TURN_LVL_METRICS into the environment
+                 if "webshop" not in cache['tag'].lower() or ("webshop" in cache['tag'].lower() and k in TURN_LVL_METRICS):
+@@ -219,7 +259,12 @@ class EnvStateManager:
+                 else:
+                     env_metric['traj_sum/' + k] = np.sum(v)
+ 
+-
++            try:
++                if 'score' in custom_metric and len(custom_metric['score']) > 0:
++                    env_metric['final_score'] = float(custom_metric['score'][-1])
++            except Exception:
++                pass
++                
+             cache['history'][-1]['metrics'] = custom_metric
+             env_metric = {f"{entry['tag']}/{k}": v for k, v in env_metric.items()}
+             cache['metrics'] = env_metric
+diff --git a/requirements.txt b/requirements.txt
+index 2fd756e..9b521bf 100644
+--- a/requirements.txt
++++ b/requirements.txt
+@@ -7,7 +7,6 @@ accelerate
+ codetiming
+ datasets
+ dill
+-flash-attn==2.7.4.post1
+ hydra-core
+ numpy
+ pandas
+@@ -15,19 +14,19 @@ pybind11
+ ray>=2.10
+ tensordict>=0.8.0,<0.9.0
+ transformers
+-vllm==0.8.2
++vllm==0.8.5
+ wandb
+ gymnasium
+ gymnasium[toy-text]
+ 
+ pyarrow>=15.0.0
+ pylatexenc
+-torchdata
++# torchdata
+ debugpy
+ 
+ together
+ anthropic
+-faiss-cpu==1.11.0
++faiss-gpu
+ 
+ # This is optional, but needs to be installed with main requirements if you want to use webshop
+ -r external/webshop-minimal/requirements.txt
+diff --git a/scripts/setup_ragen.sh b/scripts/setup_ragen.sh
+index f9a7cd9..85a93c4 100644
+--- a/scripts/setup_ragen.sh
++++ b/scripts/setup_ragen.sh
+@@ -94,10 +94,10 @@ main() {
+         pip install torch==2.5.0 --index-url https://download.pytorch.org/whl/cu124
+         
+         print_step "Installing flash-attention..."
+-        pip3 install flash-attn==2.7.4.post1 --no-build-isolation
++        # pip3 install flash-attn==2.7.4.post1 --no-build-isolation
+     else
+         print_step "Installing PyTorch without CUDA support..."
+-        pip install torch==2.5.0
++        pip install torch==2.4.0
+     fi
+     
+     # Install remaining requirements
+@@ -137,8 +137,8 @@ main() {
+     conda install conda-forge::gdown
+     mkdir -p external/webshop-minimal/webshop_minimal/data/full
+     cd external/webshop-minimal/webshop_minimal/data/full
+-    gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
+-    gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
++    # gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
++    # gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
+     cd ../../../../..
+ 
+     echo -e "${GREEN}Installation completed successfully!${NC}"
+diff --git a/train_all.sh b/train_all.sh
+index 0157306..33035f5 100755
+--- a/train_all.sh
++++ b/train_all.sh
+@@ -6,246 +6,257 @@ USE_GRPO="algorithm.adv_estimator=grpo"
+ USE_PPO="algorithm.adv_estimator=gae" # by default.
+ USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+ 
++
++# python train.py --config-name _8_2048 system.CUDA_VISIBLE_DEVICES="'0,1,2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=game_2048 $USE_PPO $USE_BASE 
++
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-1 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-1.5B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'4,5'" trainer.project_name=ragen_latest_qwen_25_3b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-3B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-3 $USE_PPO $USE_BASE 
++
+ # Section 3.1&3.2 - General Observations
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozen_lake-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozen_lake-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=bandit-ppo-multitask $USE_PPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'7'" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-ppo-frommlp $USE_PPO $USE_BASE 
+ 
+-# Section 4.1 - Filtering and critic
+-# 0.25
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-
+-wait
+-
+-# 0.5
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-
+-# 0.75
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-
+-wait
+-
+-# Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=sokoban-ppo-box1-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-wait
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.project_name=ragen_latest_qwen_25_3b_it  trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_3B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-# Section 5.1 - Reasoning Helps Generalization
++# # Section 4.1 - Filtering and critic
++# # 0.25
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
+-    agent_proxy.enable_think=False
++# # 0.5
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++
++# # 0.75
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
+-    agent_proxy.enable_think=False
++# # Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+ 
++# wait
+ 
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
++# # Section 5.1 - Reasoning Helps Generalization
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
+ 
+-# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
++#     agent_proxy.enable_think=False
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader"
+ 
+-# COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
++#     agent_proxy.enable_think=False
+ 
+-wait
+ 
+-# Section 5.2 - what leads to better reasoning?
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
++# # SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
++# # COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
+ 
++# wait
+ 
+-wait
++# # Section 5.2 - what leads to better reasoning?
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
+ 
+-# Section 6.1 varying action count
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
++# wait
+ 
+-# section 6.2 Varying prompt diversity
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
+ 
+-wait 
++# wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
++# wait
+ 
++# # Section 6.1 varying action count
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
+ 
++# # section 6.2 Varying prompt diversity
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
+ 
++# wait 
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
+ 
+-# Extension: Training 7B reasoning model
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
+ 
+-# trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
++# # Extension: Training 7B reasoning model
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
+ 
++# python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
+ 
++# # trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_PPO="algorithm.adv_estimator=gae" # by default.
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
+-# enable_response_mask: False
+-# grpo_advantage_length_weight: True
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+ 
+-# extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
++# # USE_PPO="algorithm.adv_estimator=gae" # by default.
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
+ 
+-# extension: bi-level gae
+-python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
+-    system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
+-    model_path=Qwen/Qwen2.5-0.5B-Instruct \
+-    algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
+-    agent_proxy.use_turn_scores=True \
+-    actor_rollout_ref.rollout.tp_size_check=False
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
+ 
+-# extension: webshop
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # enable_response_mask: False
++# # grpo_advantage_length_weight: True
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: bi-level gae
++# python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
++#     system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
++#     model_path=Qwen/Qwen2.5-0.5B-Instruct \
++#     algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
++#     agent_proxy.use_turn_scores=True \
++#     actor_rollout_ref.rollout.tp_size_check=False
+ 
+ 
+-# normal:sokoban
+-# extension: sokoban
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # extension: webshop
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
+ 
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++
++# # normal:sokoban
++# # extension: sokoban
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
++
++
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
++
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+ 
+ 
+-python train.py \
+-    trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+
++# python train.py \
++#     trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+

wandb/run-20251218_184253-g8kppnjh/files/diff_8d73639b99b38265453f898b8d6af7d4af50d56e.patch

@@ -0,0 +1,1162 @@
+diff --git a/config/base.yaml b/config/base.yaml
+index a1659e4..9df6250 100644
+--- a/config/base.yaml
++++ b/config/base.yaml
+@@ -9,15 +9,16 @@ seed:
+   train: 10000
+   val: 123
+ 
+-micro_batch_size_per_gpu: 4
++micro_batch_size_per_gpu: 1
+ ppo_mini_batch_size: 32
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/Qwen/Qwen2.5-0.5B-Instruct
++# Qwen/Qwen2.5-0.5B-Instruct
+ enable_response_mask: True # Enabling response mask could improve stability of rollout/old_log_prob, as P(st|history) are no longer calculated in loss here. See https://docs.google.com/document/d/1bg7obeiKTExuHHBl5uOiSpec5uLDZ2Tgvxy6li5pHX4/edit?usp=sharing for more details.
+ grpo_advantage_length_weight: False # if you do not enable this and critic/advantage_estimator is GRPO, and the critic/advantages/mean is too low, then you can try enabling this to encourage reasoning and forbid collapse
+ 
+ lora:
+   rank: 0
+-  alpha: 64
++  alpha: 16
+   target_modules: all-linear
+ 
+ actor_rollout_ref:
+@@ -46,10 +47,10 @@ actor_rollout_ref:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu} # following micro_batch_size_per_gpu
+     tensor_model_parallel_size: 1
+-    max_model_len: 3600
++    max_model_len: 7200 #3600 why** 14400
+     prompt_length: 1 # useless. Just put it here
+     response_length: 400 # single-turn response length
+-    gpu_memory_utilization: 0.5
++    gpu_memory_utilization: 0.7
+     max_num_batched_tokens: 8192 # set only when enable_chunked_prefill is true
+     temperature: 1
+     rollout_filter_ratio: 0.25
+@@ -90,26 +91,28 @@ algorithm:
+     kl_coef: 0.000
+ 
+ trainer:
+-  project_name: ragen_latest
++  project_name: 
+   experiment_name: test
+   local_log_dir: "results/"
+-  save_freq: 100
++  save_freq: -1
+   total_training_steps: 200
+   validation_steps: 1 # validation instances = validation_steps * val_env_groups * group_size
+   val_before_train: True
+   n_gpus_per_node: 1
+   test_freq: 10
+   generations_to_log_to_wandb: 
++    train: 128
+     val: 20
+   logger: [ 'console', 'wandb' ]
+   max_actor_ckpt_to_keep: 1
+   max_critic_ckpt_to_keep: 1
++  default_local_dir: /mnt/general/wanghy/RAGEN/saves/
+ 
+ agent_proxy:
+   max_context_window: -1 # set a value > 0 to enable context window for long trajectory
+-  max_turn: 5
++  max_turn: 25 #25 why** 700
+   action_sep: "||"
+-  max_actions_per_turn: 2 # how many actions can be output at most in a single turn
++  max_actions_per_turn: 1 # how many actions can be output at most in a single turn
+   use_turn_scores: False # important to GAE when applying token-level rewards to token-level advantages. If False, will take the sum of scores as the reward for the last turn.
+   enable_think: True # False -> no think RL
+   reward_normalization:
+@@ -126,11 +129,11 @@ es_manager:
+       tags: ["CoordSokoban"]
+       n_groups: [8] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+   val:
+-    env_groups: 32
+-    group_size: 16 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
++    env_groups: 256
++    group_size: 1 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
+     env_configs:
+       tags: ["CoordSokoban"]
+-      n_groups: [32] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      n_groups: [256] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ ctx_manager:
+   generation: # go to vllm
+diff --git a/config/envs.yaml b/config/envs.yaml
+index 1002ac8..ba3ac6f 100644
+--- a/config/envs.yaml
++++ b/config/envs.yaml
+@@ -2,6 +2,7 @@ custom_envs:
+   SimpleSokoban:
+     env_type: sokoban
+     max_actions_per_traj: 10 # used in environment state manager to control the actual max actions executed per trajectory
++    # ORIGNAL env_instruction
+     env_instruction: |
+       You are solving the Sokoban puzzle. 
+       You are the player and you need to push all boxes to targets. 
+@@ -12,7 +13,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+ 
+   LargerSokoban:
+@@ -65,7 +66,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+       observation_format: "grid_coord"
+ 
+@@ -111,14 +112,14 @@ custom_envs:
+ 
+   FrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: "You are solving the FrozenLake puzzle. Forbid the whole and go to the target. You may move to the unintended direction due to the slippery ice. Example answer format: <think>To forbid the hole and go to the target, I should go left then go up.</think><answer>Left || Up</answer>"
+     max_tokens: 100
+     env_config: null
+ 
+   CoordFrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: |
+       You are solving the FrozenLake puzzle. The observation includes both a symbol grid and zero-indexed coordinates for the start, goal, player, and any holes.
+       Coordinates range from the top-left corner (0, 0) to the bottom-right corner (5, 5).
+@@ -195,4 +196,80 @@ custom_envs:
+     max_actions_per_traj: 30
+     env_instruction: "You are a Lean theorem prover. Given a Lean theorem statement, propose a sequence of tactics that completes the proof. Think step by step about which tactics to apply next. Provide tactics separated by '||', for example <answer>intro || simp || rfl</answer>."
+     max_tokens: 512
+-    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
+
++    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
++
++  
++  game_2048:
++    env_type: game_2048
++    max_actions_per_traj: 700
++    env_instruction: |
++      You are playing the 2048 game on a 4x4 grid. Merge equal tiles by sliding Up, Right, Down, or Left.
++      If a move is invalid (no tiles move), a small penalty is applied. Respond with a single action.
++      Example: <answer>Up</answer>
++    max_tokens: 8192
++    env_config: null
++
++  blackjack:
++    env_type: blackjack
++    max_actions_per_traj: 10
++    env_instruction: |
++      You are playing Blackjack against a dealer. The dealer must hit on 16 or less and stand on 17 or more.
++      Choose either Stick or Hit. Respond with a single action.
++      Example: <answer>Hit</answer>
++    max_tokens: 64
++    env_config: null
++
++  rubikscube:
++    env_type: rubikscube
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.
++      Available actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.
++      - Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).
++      - Modifiers: A letter alone means 90° clockwise (e.g., 'R'). A letter with prime (') means 90° counter-clockwise (e.g., "R'").
++      Respond with a sequence of actions separated by "||".
++      Example: <answer>U</answer>
++    max_tokens: 96
++    env_config:
++      scramble_depth: 1
++      max_steps: 20
++      render_mode: "text"
++  
++  SimpleSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 2x2 box contains the numbers 1-4 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format, for example: <answer>place 1 at row 2 col 3</answer> or <answer>1,2,3</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 4
++      difficulty: "easy"
++      render_format: "with_feedback"
++      show_conflicts: false
++      show_valid_numbers: false
++      max_steps: 20
++
++  MediumSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 30
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 3x3 box contains the numbers 1-9 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format: <answer>place 5 at row 2 col 3</answer> or <answer>2,3,5</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 9
++      difficulty: "medium"
++      render_format: "with_feedback"
++      show_conflicts: true
++      show_valid_numbers: true
++      max_steps: 81
++
+diff --git a/config/eval.yaml b/config/eval.yaml
+index e65f3a5..7775fd1 100644
+--- a/config/eval.yaml
++++ b/config/eval.yaml
+@@ -1,14 +1,15 @@
+ defaults:
+   - envs
+-
++enable_response_mask: True
+ system:
+-  CUDA_VISIBLE_DEVICES: "0"
++  CUDA_VISIBLE_DEVICES: "0,1,2,3,4,5,6,7"
+ 
+ seed:
+   train: 10000
+   val: 123
+ 
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/openai/gpt-oss-20b
++# /mnt/general/share/model/tyzhu/SPA-frozenlake-qwen2.5-1.5b-instruct
+ 
+ lora:
+   rank: 0
+@@ -24,9 +25,9 @@ actor_rollout_ref:
+   rollout:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu}
+-    tensor_model_parallel_size: 1
++    tensor_model_parallel_size: 8
+     dtype: bfloat16
+-    max_model_len: 3600
++    max_model_len: 7200
+     prompt_length: 1
+     response_length: 400
+     gpu_memory_utilization: 0.9
+@@ -35,6 +36,7 @@ actor_rollout_ref:
+     free_cache_engine: True
+     enable_chunked_prefill: False
+     disable_log_stats: False
++    do_sample: True
+     val_kwargs:
+       do_sample: True
+       temperature: 0.5
+@@ -44,9 +46,9 @@ actor_rollout_ref:
+ 
+ agent_proxy:
+   max_context_window: -1
+-  max_turn: 5
++  max_turn: 25
+   action_sep: "||"
+-  max_actions_per_turn: 2
++  max_actions_per_turn: 1
+   use_turn_scores: False
+   enable_think: True
+   reward_normalization:
+@@ -59,13 +61,13 @@ es_manager:
+     env_groups: 8
+     group_size: 16
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [8]
+   val:
+     env_groups: 32
+-    group_size: 16
++    group_size: 128
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [32]
+ 
+ ctx_manager:
+diff --git a/config/evaluate_api_llm.yaml b/config/evaluate_api_llm.yaml
+index b21c95e..e2e6761 100644
+--- a/config/evaluate_api_llm.yaml
++++ b/config/evaluate_api_llm.yaml
+@@ -1,8 +1,11 @@
++#export OPENAI_BASE_URL="https://api.ohmygpt.com/v1"
++#export OPENAI_API_KEY="sk-o4sMxBkN5BB100C4D4a3T3BlBkFJF7791CA39EA14ca98041"
++#python -m ragen.eval_api hydra.searchpath='[file://./verl/verl/trainer/config]'
+ defaults:
+   - base # this is a symbolic link to the verl/verl/trainer/config/ppo_trainer.yaml file
+ 
+ model_config:
+-  model_name: gpt-4o # should be registered in model_info
++  model_name: ark-deepseek-v3-250324 # should be registered in model_info
+   max_concurrency: 16
+ 
+ model_info:
+@@ -24,9 +27,9 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+-  gpt-4o:
++  gpt-4o-mini:
+     provider_name: openai
+-    model_name: gpt-4o
++    model_name: gpt-4o-mini
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+@@ -36,21 +39,42 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  ark-deepseek-v3-250324:
++    provider_name: openai
++    model_name: ark-deepseek-v3-250324
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
+   deepseek-v3:
+     provider_name: deepseek
+     model_name: deepseek-chat
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  glm-4.6:
++    provider_name: openai
++    model_name: glm-4.6
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
++  TA/openai/gpt-oss-120b:
++    provider_name: openai
++    model_name: TA/openai/gpt-oss-120b
++    generation_kwargs:
++      temperature: 0
++      max_tokens: 8192
++      # max_retries: 5
+ 
+-
+-
++agent_proxy:
++  max_turn: 5
+ es_manager:
+   val:
+-    env_groups: 256
++    env_groups: 128
+     group_size: 1 # should be set to 1 because val temperature is set to 0 and same prompt leads to same output
+     env_configs:
+-      tags: ["CoordSokoban"]
+-      n_groups: [256] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      tags: ["rubikscube"]
++      n_groups: [128] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ 
++rollout:
++  max_model_len: 7200
+
+Submodule external/webshop-minimal contains modified content
+diff --git a/external/webshop-minimal/requirements.txt b/external/webshop-minimal/requirements.txt
+index 5a1b04f..238ed5a 100644
+--- a/external/webshop-minimal/requirements.txt
++++ b/external/webshop-minimal/requirements.txt
+@@ -4,7 +4,7 @@ flask
+ html2text
+ rank_bm25
+ pyserini
+-faiss-cpu
++faiss-gpu
+ thefuzz
+ gdown
+ spacy
+diff --git a/ragen/env/__init__.py b/ragen/env/__init__.py
+index b0f3461..bc6838e 100644
+--- a/ragen/env/__init__.py
++++ b/ragen/env/__init__.py
+@@ -12,6 +12,14 @@ from .metamathqa.env import MetaMathQAEnv
+ from .metamathqa.config import MetaMathQAEnvConfig
+ from .lean.config import LeanEnvConfig
+ from .lean.env import LeanEnv
++from .game_2048.config import Game2048EnvConfig
++from .game_2048.env import Game2048Env
++from .blackjack.config import BlackjackEnvConfig
++from .blackjack.env import BlackjackEnv
++from .rubikscube.config import RubiksCube2x2Config
++from .rubikscube.env import RubiksCube2x2Env
++from .sudoku.config import SudokuEnvConfig
++from .sudoku.env import SudokuEnv
+ 
+ 
+ REGISTERED_ENVS = {
+@@ -22,6 +30,10 @@ REGISTERED_ENVS = {
+     # 'alfworld': AlfredTXTEnv,
+     'metamathqa': MetaMathQAEnv,
+     'lean': LeanEnv,
++    'game_2048': Game2048Env,
++    'blackjack': BlackjackEnv,
++    'rubikscube': RubiksCube2x2Env,
++    'sudoku': SudokuEnv,
+ }
+ 
+ REGISTERED_ENV_CONFIGS = {
+@@ -32,6 +44,10 @@ REGISTERED_ENV_CONFIGS = {
+     # 'alfworld': AlfredEnvConfig,
+     'metamathqa': MetaMathQAEnvConfig,
+     'lean': LeanEnvConfig,
++    'game_2048': Game2048EnvConfig,
++    'blackjack': BlackjackEnvConfig,
++    'rubikscube': RubiksCube2x2Config,
++    'sudoku': SudokuEnvConfig,
+ }
+ 
+ try:
+diff --git a/ragen/env/frozen_lake/config.py b/ragen/env/frozen_lake/config.py
+index de054f4..9950c34 100644
+--- a/ragen/env/frozen_lake/config.py
++++ b/ragen/env/frozen_lake/config.py
+@@ -8,7 +8,7 @@ class FrozenLakeEnvConfig:
+     size: int = 4
+     p: float = 0.9
+     success_rate: float = 0.8
+-    is_slippery: bool = True
++    is_slippery: bool = False
+     map_seed: Optional[int] = None
+     render_mode: str = "text"
+     observation_format: str = "grid"
+diff --git a/ragen/env/frozen_lake/env.py b/ragen/env/frozen_lake/env.py
+index 9ee3add..eef11d8 100644
+--- a/ragen/env/frozen_lake/env.py
++++ b/ragen/env/frozen_lake/env.py
+@@ -13,6 +13,7 @@ from ragen.env.base import BaseDiscreteActionEnv
+ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+     def __init__(self, config: FrozenLakeEnvConfig = FrozenLakeEnvConfig()):
+         # Using mappings directly from config
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.GRID_LOOKUP = config.grid_lookup
+         self.ACTION_LOOKUP = config.action_lookup
+@@ -95,6 +96,7 @@ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+ if __name__ == "__main__":
+     import matplotlib.pyplot as plt
+     config = FrozenLakeEnvConfig(size=4, is_slippery=True)
++    # import pdb;pdb.set_trace()
+     env = FrozenLakeEnv(config)
+     print(env.reset())
+     while True:
+diff --git a/ragen/env/sokoban/env.py b/ragen/env/sokoban/env.py
+index 17cd636..6922b54 100644
+--- a/ragen/env/sokoban/env.py
++++ b/ragen/env/sokoban/env.py
+@@ -33,6 +33,7 @@ class SokobanEnv(BaseDiscreteActionEnv, GymSokobanEnv):
+     def reset(self, seed=None, mode=None):
+         try:
+             with all_seed(seed):
++                # import pdb;pdb.set_trace()
+                 self.room_fixed, self.room_state, self.box_mapping, action_sequence = generate_room(
+                     dim=self.dim_room,
+                     num_steps=self.num_gen_steps,
+diff --git a/ragen/llm_agent/agent_proxy.py b/ragen/llm_agent/agent_proxy.py
+index c25b50d..f6c4088 100644
+--- a/ragen/llm_agent/agent_proxy.py
++++ b/ragen/llm_agent/agent_proxy.py
+@@ -287,6 +287,7 @@ def main(config):
+ 	proxy = LLMAgentProxy(config, actor_wg, tokenizer)
+ 	import time
+ 	start_time = time.time()
++	# import pdb;pdb.set_trace()
+ 	rollouts = proxy.rollout(
+ 		DataProto(
+ 			batch=None,
+diff --git a/ragen/llm_agent/base_llm.py b/ragen/llm_agent/base_llm.py
+index 358a2eb..b6f7797 100644
+--- a/ragen/llm_agent/base_llm.py
++++ b/ragen/llm_agent/base_llm.py
+@@ -33,6 +33,7 @@ class OpenAIProvider(LLMProvider):
+             raise ValueError("OpenAI API key not provided and not found in environment variables")
+         
+         self.client = AsyncOpenAI(api_key=self.api_key)
++        # import pdb;pdb.set_trace()
+     
+     async def generate(self, messages: List[Dict[str, str]], **kwargs) -> LLMResponse:
+         if "o1-mini" in self.model_name:
+@@ -46,10 +47,7 @@ class OpenAIProvider(LLMProvider):
+         )
+         if response.choices[0].finish_reason in ['length', 'content_filter']:
+             raise ValueError("Content filtered or length exceeded")
+-        return LLMResponse(
+-            content=response.choices[0].message.content,
+-            model_name=response.model
+-        )
++        return LLMResponse(content=response.choices[0].message.content,model_name=response.model)
+ 
+ class DeepSeekProvider(LLMProvider):
+     """DeepSeek API provider implementation"""
+@@ -199,7 +197,7 @@ class ConcurrentLLM:
+         
+         # Queue to store unfinished or failed tasks
+         current_batch = messages_list.copy()
+-        max_retries = kwargs.get("max_retries", 100)
++        max_retries = kwargs.get("max_retries", 10)
+         retry_count = 0
+         
+         while current_batch and retry_count < max_retries:
+diff --git a/ragen/llm_agent/ctx_manager.py b/ragen/llm_agent/ctx_manager.py
+index 905247a..af20a01 100644
+--- a/ragen/llm_agent/ctx_manager.py
++++ b/ragen/llm_agent/ctx_manager.py
+@@ -88,6 +88,7 @@ class ContextManager:
+         Initialize the ContextManager.
+         Processor is used to process the image data.
+         """
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.tokenizer = tokenizer
+         self.processor = processor
+@@ -311,7 +312,7 @@ class ContextManager:
+ 
+             llm_input_texts.append(text_with_prompt)
+             messages_list.append(messages)
+-
++        # import pdb;pdb.set_trace()
+         inputs = self.tokenizer(llm_input_texts, return_tensors="pt", padding=True, padding_side="left", truncation=False) # We have truncated previously, truncation in tokenizer may cause issues.
+         input_ids, attention_mask = inputs.input_ids, inputs.attention_mask
+         position_ids = (attention_mask.cumsum(dim=-1) - 1).clamp(min=0)
+@@ -353,6 +354,30 @@ class ContextManager:
+                 key: np.sum(value) / self.env_nums[key.split("/")[0]]
+                 for key, value in metrics.items()
+             }
++            # Derived metrics for wandb logging
++            try:
++                # charts/avg_episode_return: average across all 2048 env groups
++                ep_keys = [k for k in metrics.keys() if k.endswith('/episodic_return')]
++                if len(ep_keys) > 0:
++                    ep_vals = []
++                    for k in ep_keys:
++                        tag = k.split('/')[0]
++                        denom = self.env_nums.get(tag, max(1, len(metrics[k])))
++                        ep_vals.append(float(np.sum(metrics[k]) / denom))
++                    mean_metrics["charts/avg_episode_return"] = float(np.mean(ep_vals))
++            except Exception:
++                pass
++            try:
++                # rollout/max_tile: max over all envs in this batch
++                tile_keys = [k for k in metrics.keys() if k.endswith('/max_tile')]
++                if len(tile_keys) > 0:
++                    tile_vals = []
++                    for k in tile_keys:
++                        tile_vals.extend(metrics[k])
++                    if len(tile_vals) > 0:
++                        mean_metrics["rollout/max_tile"] = int(np.max(tile_vals))
++            except Exception:
++                pass
+             for key, values in metrics.items():
+                 if not isinstance(values, list):
+                     continue
+diff --git a/ragen/llm_agent/es_manager.py b/ragen/llm_agent/es_manager.py
+index b87a2b3..1a49d3e 100644
+--- a/ragen/llm_agent/es_manager.py
++++ b/ragen/llm_agent/es_manager.py
+@@ -128,18 +128,40 @@ class EnvStateManager:
+         env_outputs: List[Dict]
+             {env_id: int, history: List[Dict][{state: str, actions: List[str], reward: float, info: Dict, llm_response: str, llm_raw_response: str, (Optional)images: List[PIL.Image.Image]}]}
+         """
++        # def _execute_actions(env, actions):
++        #     acc_reward, turn_info, turn_done = 0, {}, False
++        #     executed_actions = []
++        #     for action in actions:
++        #         _, reward, done, info = env.step(action)
++        #         acc_reward += reward
++        #         turn_info.update(info) # NOTE: currently use last info for multi-action
++        #         executed_actions.append(action)
++        #         if done:
++        #             turn_done = True
++        #             break
++            
++        #     return acc_reward, turn_info, turn_done, executed_actions
+         def _execute_actions(env, actions):
+-            acc_reward, turn_info, turn_done = 0, {}, False
++            acc_reward, turn_info, turn_done = 0.0, {}, False
++            raw_acc_reward = 0.0
+             executed_actions = []
+             for action in actions:
+                 _, reward, done, info = env.step(action)
+-                acc_reward += reward
++                acc_reward += float(reward)
++                try:
++                    raw_acc_reward += float(info.get('raw_reward', 0.0))
++                except Exception:
++                    pass
+                 turn_info.update(info) # NOTE: currently use last info for multi-action
+                 executed_actions.append(action)
+                 if done:
+                     turn_done = True
+                     break
+-            
++            # Overwrite per-turn raw_reward to reflect the sum across all executed actions in this turn
++            try:
++                turn_info['raw_reward'] = float(raw_acc_reward)
++            except Exception:
++                pass
+             return acc_reward, turn_info, turn_done, executed_actions
+ 
+         def _log_env_state(status, history, cur_obs, max_actions_per_traj, executed_actions, all_actions, acc_reward, turn_done, turn_info, env_input):
+@@ -198,6 +220,18 @@ class EnvStateManager:
+                 'success': float(status.terminated and (not status.truncated)),
+                 'num_actions': status.num_actions,
+             }
++            # Add episodic-level metrics
++            # try:
++            #     # Sum of per-turn rewards equals the episodic return (env-shaped reward)
++            #     env_metric['episodic_return'] = float(sum(status.rewards))
++            # except Exception:
++            #     pass
++            try:
++                # Final max tile on the board at the end of the rollout
++                import numpy as _np
++                env_metric['max_tile'] = int(_np.max(entry['env'].grid))
++            except Exception:
++                pass
+             custom_metric = {}
+             for turn in cache['history']:
+                 for k, v in turn.get('info', {}).items():
+@@ -212,6 +246,12 @@ class EnvStateManager:
+                             "Skipping non-numeric metric '%s' with value %r for env %s.",
+                             k, v, entry['tag']
+                         )
++            # Add episodic_return as the SUM of raw_reward across steps (align with CleanRL)
++            try:
++                if 'raw_reward' in custom_metric:
++                    env_metric['episodic_return'] = float(np.sum(custom_metric['raw_reward']))
++            except Exception:
++                pass
+             for k, v in custom_metric.items():
+                 # TODO: Move TURN_LVL_METRICS into the environment
+                 if "webshop" not in cache['tag'].lower() or ("webshop" in cache['tag'].lower() and k in TURN_LVL_METRICS):
+@@ -219,7 +259,12 @@ class EnvStateManager:
+                 else:
+                     env_metric['traj_sum/' + k] = np.sum(v)
+ 
+-
++            try:
++                if 'score' in custom_metric and len(custom_metric['score']) > 0:
++                    env_metric['final_score'] = float(custom_metric['score'][-1])
++            except Exception:
++                pass
++                
+             cache['history'][-1]['metrics'] = custom_metric
+             env_metric = {f"{entry['tag']}/{k}": v for k, v in env_metric.items()}
+             cache['metrics'] = env_metric
+diff --git a/requirements.txt b/requirements.txt
+index 2fd756e..9b521bf 100644
+--- a/requirements.txt
++++ b/requirements.txt
+@@ -7,7 +7,6 @@ accelerate
+ codetiming
+ datasets
+ dill
+-flash-attn==2.7.4.post1
+ hydra-core
+ numpy
+ pandas
+@@ -15,19 +14,19 @@ pybind11
+ ray>=2.10
+ tensordict>=0.8.0,<0.9.0
+ transformers
+-vllm==0.8.2
++vllm==0.8.5
+ wandb
+ gymnasium
+ gymnasium[toy-text]
+ 
+ pyarrow>=15.0.0
+ pylatexenc
+-torchdata
++# torchdata
+ debugpy
+ 
+ together
+ anthropic
+-faiss-cpu==1.11.0
++faiss-gpu
+ 
+ # This is optional, but needs to be installed with main requirements if you want to use webshop
+ -r external/webshop-minimal/requirements.txt
+diff --git a/scripts/setup_ragen.sh b/scripts/setup_ragen.sh
+index f9a7cd9..85a93c4 100644
+--- a/scripts/setup_ragen.sh
++++ b/scripts/setup_ragen.sh
+@@ -94,10 +94,10 @@ main() {
+         pip install torch==2.5.0 --index-url https://download.pytorch.org/whl/cu124
+         
+         print_step "Installing flash-attention..."
+-        pip3 install flash-attn==2.7.4.post1 --no-build-isolation
++        # pip3 install flash-attn==2.7.4.post1 --no-build-isolation
+     else
+         print_step "Installing PyTorch without CUDA support..."
+-        pip install torch==2.5.0
++        pip install torch==2.4.0
+     fi
+     
+     # Install remaining requirements
+@@ -137,8 +137,8 @@ main() {
+     conda install conda-forge::gdown
+     mkdir -p external/webshop-minimal/webshop_minimal/data/full
+     cd external/webshop-minimal/webshop_minimal/data/full
+-    gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
+-    gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
++    # gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
++    # gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
+     cd ../../../../..
+ 
+     echo -e "${GREEN}Installation completed successfully!${NC}"
+diff --git a/train_all.sh b/train_all.sh
+index 0157306..33035f5 100755
+--- a/train_all.sh
++++ b/train_all.sh
+@@ -6,246 +6,257 @@ USE_GRPO="algorithm.adv_estimator=grpo"
+ USE_PPO="algorithm.adv_estimator=gae" # by default.
+ USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+ 
++
++# python train.py --config-name _8_2048 system.CUDA_VISIBLE_DEVICES="'0,1,2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=game_2048 $USE_PPO $USE_BASE 
++
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-1 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-1.5B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'4,5'" trainer.project_name=ragen_latest_qwen_25_3b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-3B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-3 $USE_PPO $USE_BASE 
++
+ # Section 3.1&3.2 - General Observations
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozen_lake-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozen_lake-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=bandit-ppo-multitask $USE_PPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'7'" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-ppo-frommlp $USE_PPO $USE_BASE 
+ 
+-# Section 4.1 - Filtering and critic
+-# 0.25
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-
+-wait
+-
+-# 0.5
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-
+-# 0.75
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-
+-wait
+-
+-# Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=sokoban-ppo-box1-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-wait
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.project_name=ragen_latest_qwen_25_3b_it  trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_3B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-# Section 5.1 - Reasoning Helps Generalization
++# # Section 4.1 - Filtering and critic
++# # 0.25
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
+-    agent_proxy.enable_think=False
++# # 0.5
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++
++# # 0.75
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
+-    agent_proxy.enable_think=False
++# # Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+ 
++# wait
+ 
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
++# # Section 5.1 - Reasoning Helps Generalization
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
+ 
+-# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
++#     agent_proxy.enable_think=False
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader"
+ 
+-# COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
++#     agent_proxy.enable_think=False
+ 
+-wait
+ 
+-# Section 5.2 - what leads to better reasoning?
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
++# # SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
++# # COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
+ 
++# wait
+ 
+-wait
++# # Section 5.2 - what leads to better reasoning?
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
+ 
+-# Section 6.1 varying action count
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
++# wait
+ 
+-# section 6.2 Varying prompt diversity
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
+ 
+-wait 
++# wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
++# wait
+ 
++# # Section 6.1 varying action count
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
+ 
++# # section 6.2 Varying prompt diversity
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
+ 
++# wait 
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
+ 
+-# Extension: Training 7B reasoning model
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
+ 
+-# trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
++# # Extension: Training 7B reasoning model
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
+ 
++# python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
+ 
++# # trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_PPO="algorithm.adv_estimator=gae" # by default.
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
+-# enable_response_mask: False
+-# grpo_advantage_length_weight: True
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+ 
+-# extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
++# # USE_PPO="algorithm.adv_estimator=gae" # by default.
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
+ 
+-# extension: bi-level gae
+-python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
+-    system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
+-    model_path=Qwen/Qwen2.5-0.5B-Instruct \
+-    algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
+-    agent_proxy.use_turn_scores=True \
+-    actor_rollout_ref.rollout.tp_size_check=False
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
+ 
+-# extension: webshop
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # enable_response_mask: False
++# # grpo_advantage_length_weight: True
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: bi-level gae
++# python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
++#     system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
++#     model_path=Qwen/Qwen2.5-0.5B-Instruct \
++#     algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
++#     agent_proxy.use_turn_scores=True \
++#     actor_rollout_ref.rollout.tp_size_check=False
+ 
+ 
+-# normal:sokoban
+-# extension: sokoban
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # extension: webshop
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
+ 
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++
++# # normal:sokoban
++# # extension: sokoban
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
++
++
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
++
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+ 
+ 
+-python train.py \
+-    trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+
++# python train.py \
++#     trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+

wandb/run-20251218_184253-g8kppnjh/files/requirements.txt

@@ -0,0 +1,315 @@
+colorama==0.4.6
+psutil==7.1.2
+ragen==0.1
+ragen==0.1
+setuptools==80.9.0
+wheel==0.45.1
+pip==25.3
+ragen==0.1
+pytz==2025.2
+nvidia-cusparselt-cu12==0.6.2
+mpmath==1.3.0
+typing_extensions==4.15.0
+sympy==1.13.1
+pylatexenc==2.10
+nvidia-cusolver-cu12==11.6.1.9
+numpy==1.26.4
+Markdown==3.10
+absl-py==2.3.1
+pyasn1_modules==0.4.2
+Werkzeug==3.1.3
+networkx==3.3
+MarkupSafe==2.1.5
+fsspec==2025.9.0
+filelock==3.19.1
+nvidia-cudnn-cu12==9.1.0.70
+Jinja2==3.1.6
+tensorboard==2.20.0
+triton==3.2.0
+pillow==11.3.0
+xxhash==3.6.0
+nvidia-nvtx-cu12==12.4.127
+nvidia-nvjitlink-cu12==12.4.127
+nvidia-nccl-cu12==2.21.5
+nvidia-curand-cu12==10.3.5.147
+wrapt==2.0.1
+nvidia-cufft-cu12==11.2.1.3
+pyjnius==1.7.0
+nvidia-cuda-runtime-cu12==12.4.127
+nvidia-cuda-nvrtc-cu12==12.4.127
+websockets==15.0.1
+nvidia-cuda-cupti-cu12==12.4.127
+zipp==3.23.0
+nvidia-cublas-cu12==12.4.5.8
+nvidia-cusparse-cu12==12.3.1.170
+flash_attn==2.8.3
+rsa==4.9.1
+einops==0.8.1
+py-cpuinfo==9.0.0
+pure_eval==0.2.3
+ptyprocess==0.7.0
+gym-notices==0.1.0
+flatbuffers==25.9.23
+fastrlock==0.8.3
+Farama-Notifications==0.0.4
+faiss-gpu==1.7.2
+antlr4-python3-runtime==4.9.3
+wcwidth==0.2.14
+wasabi==1.1.3
+uvloop==0.22.1
+urllib3==2.5.0
+tzdata==2025.2
+typing-inspection==0.4.2
+traitlets==5.14.3
+tqdm==4.67.1
+tomli==2.3.0
+threadpoolctl==3.6.0
+tenacity==9.1.2
+tabulate==0.9.0
+spacy-loggers==1.0.5
+spacy-legacy==3.0.12
+soupsieve==2.8
+sniffio==1.3.1
+smmap==5.0.2
+six==1.17.0
+shellingham==1.5.4
+sentencepiece==0.2.1
+scipy==1.15.3
+safetensors==0.6.2
+rpds-py==0.29.0
+rignore==0.7.6
+regex==2025.11.3
+RapidFuzz==3.14.3
+rank-bm25==0.2.2
+pyzmq==27.1.0
+PyYAML==6.0.3
+pytokens==0.3.0
+python-multipart==0.0.20
+python-json-logger==4.0.0
+python-dotenv==1.2.1
+PySocks==1.7.1
+pyparsing==3.2.5
+PyJWT==2.10.1
+Pygments==2.19.2
+pygame==2.6.1
+pydantic_core==2.41.5
+pycparser==2.23
+pycountry==24.6.1
+pybind11==3.0.1
+pyarrow==22.0.0
+psutil==7.1.3
+protobuf==4.25.8
+propcache==0.4.1
+prometheus_client==0.23.1
+platformdirs==4.5.0
+pexpect==4.9.0
+pathspec==0.12.1
+partial-json-parser==0.2.1.1.post6
+parso==0.8.5
+packaging==25.0
+orjson==3.11.4
+opentelemetry-semantic-conventions-ai==0.4.13
+opencv-python-headless==4.12.0.88
+ninja==1.13.0
+nest-asyncio==1.6.0
+mypy_extensions==1.1.0
+murmurhash==1.0.15
+multidict==6.7.0
+msgspec==0.19.0
+msgpack==1.1.2
+mdurl==0.1.2
+loguru==0.7.3
+llvmlite==0.44.0
+llguidance==0.7.30
+lark==1.2.2
+kiwisolver==1.4.9
+joblib==1.5.2
+jiter==0.12.0
+itsdangerous==2.2.0
+interegular==0.3.3
+ImageIO==2.37.2
+idna==3.11
+humanfriendly==10.0
+httpx-sse==0.4.3
+httptools==0.7.1
+html2text==2025.4.15
+hf-xet==1.2.0
+h11==0.16.0
+grpcio==1.76.0
+frozenlist==1.8.0
+fonttools==4.60.1
+executing==2.2.1
+exceptiongroup==1.3.0
+eval_type_backport==0.2.2
+docstring_parser==0.17.0
+dnspython==2.8.0
+distro==1.9.0
+diskcache==5.6.3
+dill==0.4.0
+decorator==5.2.1
+debugpy==1.8.17
+Cython==3.2.1
+cymem==2.0.13
+cycler==0.12.1
+cupy-cuda12x==13.6.0
+contourpy==1.3.2
+colorama==0.4.6
+codetiming==1.4.0
+cloudpickle==3.1.2
+cloudpathlib==0.23.0
+click==8.3.1
+charset-normalizer==3.4.4
+certifi==2025.11.12
+catalogue==2.0.10
+cachetools==6.2.2
+blis==1.3.2
+blinker==1.9.0
+blake3==1.0.8
+attrs==25.4.0
+async-timeout==5.0.1
+asttokens==3.0.1
+astor==0.8.1
+annotated-types==0.7.0
+annotated-doc==0.0.4
+airportsdata==20250909
+aiohappyeyeballs==2.6.1
+yarl==1.22.0
+uvicorn==0.38.0
+typer-slim==0.20.0
+thefuzz==0.22.1
+stack-data==0.6.3
+srsly==2.5.1
+smart_open==7.5.0
+sentry-sdk==2.44.0
+scikit-learn==1.7.2
+requests==2.32.5
+referencing==0.37.0
+python-dateutil==2.9.0.post0
+pydantic==2.12.4
+prompt_toolkit==3.0.52
+preshed==3.0.11
+opentelemetry-proto==1.26.0
+omegaconf==2.3.0
+numba==0.61.2
+nltk==3.9.2
+multiprocess==0.70.18
+matplotlib-inline==0.2.1
+markdown-it-py==4.0.0
+jedi==0.19.2
+importlib_metadata==8.0.0
+httpcore==1.0.9
+gymnasium==1.2.2
+gym==0.26.2
+googleapis-common-protos==1.72.0
+gitdb==4.0.12
+gguf==0.17.1
+Flask==3.1.2
+faiss-cpu==1.13.0
+email-validator==2.3.0
+depyf==0.18.0
+Deprecated==1.3.1
+coloredlogs==15.0.1
+cffi==2.0.0
+black==25.11.0
+beautifulsoup4==4.14.2
+anyio==4.11.0
+aiosignal==1.4.0
+watchfiles==1.1.1
+tiktoken==0.12.0
+starlette==0.49.3
+sse-starlette==3.0.3
+rich==14.2.0
+pydantic-settings==2.12.0
+pydantic-extra-types==2.10.6
+pandas==2.3.3
+opentelemetry-exporter-otlp-proto-common==1.26.0
+onnxruntime==1.23.2
+matplotlib==3.10.7
+lm-format-enforcer==0.10.12
+jsonschema-specifications==2025.9.1
+ipython==8.37.0
+hydra-core==1.3.2
+huggingface-hub==0.36.0
+httpx==0.28.1
+gym-sokoban==0.0.6
+GitPython==3.1.45
+cryptography==46.0.3
+confection==0.1.5
+cleantext==1.1.4
+aiohttp==3.13.2
+weasel==0.4.3
+wandb==0.23.0
+typer==0.19.2
+virtualenv==20.35.4
+pyasn1==0.6.1
+torch==2.6.0
+tokenizers==0.22.1
+thinc==8.3.9
+rich-toolkit==0.15.1
+prometheus-fastapi-instrumentator==7.1.0
+openai==2.8.0
+jsonschema==4.25.1
+gdown==5.2.0
+fastapi==0.121.2
+anthropic==0.73.0
+xformers==0.0.29.post2
+transformers==4.57.1
+proto-plus==1.26.1
+torchvision==0.21.0
+torchdata==0.11.0
+torchaudio==2.6.0
+together==1.5.30
+tensordict==0.8.3
+spacy==3.8.9
+ray==2.51.1
+outlines_core==0.1.26
+mistral_common==1.8.5
+mcp==1.21.1
+fastapi-cloud-cli==0.3.1
+fastapi-cli==0.0.16
+datasets==4.4.1
+accelerate==1.11.0
+xgrammar==0.1.18
+pyserini==1.2.0
+peft==0.18.0
+outlines==0.1.11
+opentelemetry-exporter-otlp-proto-http==1.26.0
+opentelemetry-exporter-otlp-proto-grpc==1.26.0
+kimina-client==0.2.1
+compressed-tensors==0.9.3
+opentelemetry-exporter-otlp==1.26.0
+vllm==0.8.5
+py-spy==0.4.1
+opencensus-context==0.1.3
+distlib==0.4.0
+colorful==0.5.8
+tensorboard-data-server==0.7.2
+google-auth==2.43.0
+google-api-core==2.28.1
+aiohttp-cors==0.8.1
+opentelemetry-exporter-prometheus==0.59b0
+opencensus==0.11.4
+verl==0.5.0.dev0
+opentelemetry-api==1.38.0
+opentelemetry-semantic-conventions==0.59b0
+opentelemetry-sdk==1.38.0
+typeguard==4.4.4
+shtab==1.8.0
+tyro==0.9.35
+verl==0.5.0.dev0
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2

wandb/run-20251218_184253-g8kppnjh/files/wandb-metadata.json

@@ -0,0 +1,92 @@
+{
+  "os": "Linux-5.14.0-284.25.1.el9_2.x86_64-x86_64-with-glibc2.35",
+  "python": "CPython 3.10.19",
+  "startedAt": "2025-12-18T10:42:53.636495Z",
+  "args": [
+    "--track"
+  ],
+  "program": "/mnt/general/wanghy/RAGEN/cleanrl/cleanrl/ppo_rubikscube.py",
+  "codePath": "cleanrl/cleanrl/ppo_rubikscube.py",
+  "codePathLocal": "cleanrl/cleanrl/ppo_rubikscube.py",
+  "git": {
+    "remote": "https://github.com/mll-lab-nu/RAGEN.git",
+    "commit": "8d73639b99b38265453f898b8d6af7d4af50d56e"
+  },
+  "email": "haoyu-wa22@mails.tsinghua.edu.cn",
+  "root": "/mnt/general/wanghy/RAGEN",
+  "host": "pt-4d654cf4576f4d23ad3d3919f12932fe-worker-0",
+  "executable": "/root/local/miniconda3/envs/ragen/bin/python",
+  "cpu_count": 64,
+  "cpu_count_logical": 128,
+  "gpu": "NVIDIA H100 80GB HBM3",
+  "gpu_count": 8,
+  "disk": {
+    "/": {
+      "total": "5153960755200",
+      "used": "447529222144"
+    }
+  },
+  "memory": {
+    "total": "2163642122240"
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-35e2d43d-4067-82ce-90d4-def9e389bf28"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-af4135e3-88f2-e9ac-518d-502c75a85429"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-d7fdeeba-fe9b-ec03-d9f7-6724fe4266b5"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-ccc4f668-3882-5a8e-2c07-c5cd08f6f666"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-7b73c0cf-d3d5-e10c-7176-a43be1e41001"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-81b58d94-5d1f-8ec2-f9d2-fd56172ed177"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-03e8bc66-3b44-6794-49fd-5392fbdda6d1"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-b7cf0ec6-7c29-1179-dceb-09565da51890"
+    }
+  ],
+  "cudaVersion": "12.4",
+  "writerId": "xtmqsnhbpijxn98k1keqte2hruuy3myj"
+}

wandb/run-20251218_184253-g8kppnjh/files/wandb-summary.json

@@ -0,0 +1 @@
+{"train/policy_loss":-7.938966155052185e-05,"losses/value_loss":0.011233147233724594,"_runtime":398.070420408,"charts/train_success_rate":0.9559471365638766,"train/learning_rate":2.561475409836034e-07,"losses/policy_loss":-7.938966155052185e-05,"_step":999424,"_timestamp":1.766054973203551e+09,"train/success_rate":0.9559471365638766,"eval/success_rate":0.9395,"train/clipfrac":0,"rollout/ep_rew_mean":1,"charts/SPS":2564,"eval/episodes":4000,"perf/SPS":2564,"_wandb":{"runtime":398},"train/entropy":0.6522258520126343,"eval/avg_return":0.9395,"train/approx_kl":1.0943040251731873e-07,"train/old_approx_kl":-3.494042903184891e-05,"train/value_loss":0.011233147233724594,"charts/avg_reward":0.2119140625,"eval/std_return":0.23841088481862568,"rollout/success_rate":1,"charts/avg_value":0.7838361263275146,"charts/progress":100,"global_step":999424,"rollout/ep_len_mean":2,"losses/explained_variance":0.8802414909005165}

wandb/run-20251218_184253-g8kppnjh/run-g8kppnjh.wandb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:acc5f34010aa9cd646f8ad9a1d408d705346725e08d9f5548af728cf52e903db
+size 47336006

wandb/run-20251218_185248-ts36d12x/files/code/cleanrl/cleanrl/ppo_rubikscube.py

@@ -0,0 +1,517 @@
+# PPO with small MLP for RAGEN Rubik's Cube 2x2 using the existing env (no env edits)
+import os
+import random
+import time
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Tuple, Dict, Any, List
+import json
+import re
+
+import gymnasium as gym
+import numpy as np
+import torch
+import torch.nn as nn
+import torch.optim as optim
+import tyro
+from torch.distributions.categorical import Categorical
+
+import sys
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), '../../'))
+
+from ragen.env.rubikscube.env import RubiksCube2x2Env
+from ragen.env.rubikscube.config import RubiksCube2x2Config
+
+
+class RubiksCubeWrapper(gym.Env):
+    """
+    Adapter to use ragen RubiksCube2x2Env with Gymnasium vector API.
+    - Converts text observation to one-hot vector of 24 stickers x 6 colors.
+    - Maps agent actions [0..11] to env actions [1..12].
+    - Exposes proper observation_space and action_space.
+    """
+    metadata = {"render_modes": ["rgb_array", "human", "ansi"]}
+
+    def __init__(self, env: RubiksCube2x2Env):
+        super().__init__()
+        self._env = env
+        # 24 stickers, 6 colors -> one-hot size 144
+        self._colors = ['W', 'O', 'G', 'R', 'B', 'Y']
+        self._color_to_idx = {c: i for i, c in enumerate(self._colors)}
+        self.observation_space = gym.spaces.Box(low=0.0, high=1.0, shape=(24 * len(self._colors),), dtype=np.float32)
+        self.action_space = gym.spaces.Discrete(12)
+        # precompile regex to extract 4 letters within [X, X]\n [X, X]
+        # Lines look like: "Up (U):    [W, W]\n       [W, W]"
+        self._face_pat = re.compile(r"\[(?:\s*([A-Z])\s*,\s*([A-Z])\s*\])\n\s*\[(?:\s*([A-Z])\s*,\s*([A-Z])\s*\])")
+
+    def _encode_obs(self, text_obs: str) -> np.ndarray:
+        # Extract the six face blocks in the order U, L, F, R, B, D as rendered
+        faces_order = ["Up (U):", "Left (L):", "Front (F):", "Right (R):", "Back (B):", "Down (D):"]
+        onehots: List[int] = []
+        # Build a map from header to its block text
+        lines = text_obs.splitlines()
+        # Collect blocks starting after header line and including the next line for second row
+        i = 0
+        blocks: List[str] = []
+        while i < len(lines):
+            line = lines[i]
+            for header in faces_order:
+                if line.startswith(header):
+                    # Join current line's bracketed pair and the next line (which contains the second pair)
+                    # Remove the "Header:" prefix to keep only the bracket portions
+                    content = line[len(header):].strip()
+                    next_line = lines[i + 1] if i + 1 < len(lines) else ""
+                    block = f"{content}\n{next_line}"
+                    blocks.append(block)
+                    break
+            i += 1
+        # Fallback: if regex fails, return zeros
+        if len(blocks) != 6:
+            return np.zeros(24 * len(self._colors), dtype=np.float32)
+        stickers: List[int] = []
+        for blk in blocks:
+            m = self._face_pat.search(blk)
+            if not m:
+                return np.zeros(24 * len(self._colors), dtype=np.float32)
+            # order: 0,1,2,3 per render() doc
+            c0, c1, c2, c3 = m.group(1), m.group(2), m.group(3), m.group(4)
+            stickers.extend([c0, c1, c2, c3])
+        # stickers now length 24; convert to one-hot
+        grid = np.zeros((24, len(self._colors)), dtype=np.float32)
+        for idx, ch in enumerate(stickers):
+            cidx = self._color_to_idx.get(ch, None)
+            if cidx is not None:
+                grid[idx, cidx] = 1.0
+        return grid.reshape(-1)
+
+    def reset(self, *, seed: int | None = None, options: Dict[str, Any] | None = None):
+        text_obs = self._env.reset(seed=seed)
+        obs = self._encode_obs(text_obs)
+        return obs, {}
+
+    def step(self, action: int):
+        mapped = int(action) + 1  # 0..11 -> 1..12
+        text_obs, reward, done, info = self._env.step(mapped)
+        obs = self._encode_obs(text_obs)
+        terminated = bool(done)
+        truncated = False
+        return obs, float(reward), terminated, truncated, info or {}
+
+    def render(self):
+        return self._env.render()
+
+    def close(self):
+        self._env.close()
+
+
+@dataclass
+class Args:
+    exp_name: str = os.path.basename(__file__)[: -len(".py")]
+    seed: int = 1
+    torch_deterministic: bool = True
+    cuda: bool = True
+    track: bool = True
+    wandb_project_name: str = "cleanRL"
+    wandb_entity: str | None = None
+    capture_video: bool = False
+
+    # Algorithm
+    env_id: str = "RubiksCube2x2"
+    total_timesteps: int = 1000_000
+    learning_rate: float = 2.5e-4
+    num_envs: int = 8
+    num_steps: int = 128
+    anneal_lr: bool = True
+    gamma: float = 0.99
+    gae_lambda: float = 0.95
+    num_minibatches: int = 4
+    update_epochs: int = 4
+    norm_adv: bool = True
+    clip_coef: float = 0.2
+    clip_vloss: bool = True
+    ent_coef: float = 0.01
+    vf_coef: float = 0.5
+    max_grad_norm: float = 0.5
+    target_kl: float | None = None
+
+    # Rubik specific
+    scramble_depth: int = 1
+    max_steps_env: int = 20
+
+    # runtime filled
+    batch_size: int = 0
+    minibatch_size: int = 0
+    num_iterations: int = 0
+    eval_splits: int = 2
+    eval_episodes: int = 4000
+
+
+def make_env(idx, run_name, seed, scramble_depth, max_steps_env, capture_video=False):
+    def thunk():
+        config = RubiksCube2x2Config(scramble_depth=scramble_depth, max_steps=max_steps_env, render_mode='text')
+        env = RubiksCube2x2Env(config)
+        env = RubiksCubeWrapper(env)
+        env = gym.wrappers.TimeLimit(env, max_episode_steps=max_steps_env)
+        env = gym.wrappers.RecordEpisodeStatistics(env)
+        if capture_video and idx == 0:
+            env = gym.wrappers.RecordVideo(env, f"videos/{run_name}")
+        return env
+    return thunk
+
+
+def layer_init(layer, std=np.sqrt(2), bias_const=0.0):
+    torch.nn.init.orthogonal_(layer.weight, std)
+    torch.nn.init.constant_(layer.bias, bias_const)
+    return layer
+
+
+class Agent(nn.Module):
+    def __init__(self, envs):
+        super().__init__()
+        obs_shape = int(np.array(envs.single_observation_space.shape).prod())
+        hidden = 128
+        self.critic = nn.Sequential(
+            layer_init(nn.Linear(obs_shape, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, 1), std=1.0),
+        )
+        self.actor = nn.Sequential(
+            layer_init(nn.Linear(obs_shape, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, hidden)),
+            nn.Tanh(),
+            layer_init(nn.Linear(hidden, envs.single_action_space.n), std=0.01),
+        )
+
+    def get_value(self, x):
+        return self.critic(x)
+
+    def get_action_and_value(self, x, action=None):
+        logits = self.actor(x)
+        probs = Categorical(logits=logits)
+        if action is None:
+            action = probs.sample()
+        return action, probs.log_prob(action), probs.entropy(), self.critic(x)
+
+
+if __name__ == "__main__":
+    args = tyro.cli(Args)
+    args.batch_size = int(args.num_envs * args.num_steps)
+    args.minibatch_size = int(args.batch_size // args.num_minibatches)
+    args.num_iterations = args.total_timesteps // args.batch_size
+    run_name = f"{args.env_id}__{args.exp_name}__{args.seed}__{int(time.time())}"
+
+    if args.track:
+        import wandb
+        wandb.init(
+            project=args.wandb_project_name,
+            entity=args.wandb_entity,
+            config=vars(args),
+            name=run_name,
+            monitor_gym=True,
+            save_code=True,
+        )
+        try:
+            wandb.define_metric("global_step")
+            for prefix in ["train/*", "rollout/*", "eval/*", "losses/*", "charts/*", "perf/*"]:
+                wandb.define_metric(prefix, step_metric="global_step")
+        except Exception:
+            pass
+
+    # seeding
+    random.seed(args.seed)
+    np.random.seed(args.seed)
+    torch.manual_seed(args.seed)
+    torch.backends.cudnn.deterministic = args.torch_deterministic
+
+    device = torch.device("cuda" if torch.cuda.is_available() and args.cuda else "cpu")
+
+    # envs
+    envs = gym.vector.SyncVectorEnv([
+        make_env(i, run_name, args.seed, args.scramble_depth, args.max_steps_env, args.capture_video)
+        for i in range(args.num_envs)
+    ])
+    assert isinstance(envs.single_action_space, gym.spaces.Discrete)
+
+    agent = Agent(envs).to(device)
+    optimizer = optim.Adam(agent.parameters(), lr=args.learning_rate, eps=1e-5)
+
+    # storage
+    obs = torch.zeros((args.num_steps, args.num_envs) + envs.single_observation_space.shape).to(device)
+    actions = torch.zeros((args.num_steps, args.num_envs) + envs.single_action_space.shape).to(device)
+    logprobs = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    rewards = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    dones = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    values = torch.zeros((args.num_steps, args.num_envs)).to(device)
+
+    # start
+    global_step = 0
+    start_time = time.time()
+    next_obs, _ = envs.reset(seed=args.seed)
+    next_obs = torch.Tensor(next_obs).to(device)
+    next_done = torch.zeros(args.num_envs).to(device)
+
+    # eval helper similar to FrozenLake: collect greedy eval trajectories and write metrics.json
+    def collect_eval_trajectories(agent_model, make_env_fn, n_episodes, step_tag):
+        out_dir = Path(f"runs/{run_name}/trajectories/step_{step_tag}")
+        out_dir.mkdir(parents=True, exist_ok=True)
+        out_path = out_dir / "trajectories.jsonl"
+        env = make_env_fn()
+        collected = 0
+        summary_returns = []
+        summary_success = []
+        with out_path.open("w") as f:
+            while collected < n_episodes:
+                state, _ = env.reset(seed=args.seed + 100000 + collected)
+                traj_states = [state.tolist()]
+                traj_actions = []
+                traj_rewards = []
+                traj_dones = []
+                traj_success = []
+                done = False
+                step_count = 0
+                max_eval_steps = getattr(env, '_max_episode_steps', None) or int(args.max_steps_env)
+                while not done:
+                    with torch.no_grad():
+                        logits = agent_model.actor(torch.tensor(state, dtype=torch.float32, device=device).unsqueeze(0))
+                        action = int(torch.argmax(logits, dim=1).item())
+                    next_state, reward, terminated, truncated, info = env.step(action)
+                    traj_actions.append(int(action))
+                    traj_rewards.append(float(reward))
+                    step_count += 1
+                    d = bool(terminated) or bool(truncated) or (step_count >= max_eval_steps)
+                    traj_dones.append(d)
+                    traj_success.append(bool((info or {}).get('success', False)))
+                    state = next_state
+                    traj_states.append(state.tolist())
+                    done = d
+                ep_ret = float(sum(traj_rewards))
+                ep_succ = bool(any(traj_success))
+                record = {
+                    "states": traj_states,
+                    "actions": traj_actions,
+                    "rewards": traj_rewards,
+                    "dones": traj_dones,
+                    "success": traj_success,
+                    "episode_return": ep_ret,
+                    "episode_success": ep_succ,
+                }
+                f.write(json.dumps(record) + "\n")
+                collected += 1
+                summary_returns.append(ep_ret)
+                summary_success.append(1.0 if ep_succ else 0.0)
+        env.close()
+        try:
+            metrics = {
+                "global_step": int(step_tag),
+                "episodes": int(n_episodes),
+                "success_rate": float(np.mean(summary_success)) if len(summary_success) else 0.0,
+                "avg_return": float(np.mean(summary_returns)) if len(summary_returns) else 0.0,
+                "std_return": float(np.std(summary_returns)) if len(summary_returns) else 0.0,
+            }
+            with (out_dir / "metrics.json").open("w") as mf:
+                json.dump(metrics, mf)
+        except Exception as e:
+            print(f"Warning: failed to write eval metrics: {e}")
+
+    # training loop
+    eval_every_iters = max(1, args.num_iterations // args.eval_splits)
+    for iteration in range(1, args.num_iterations + 1):
+        # Anneal LR
+        if args.anneal_lr:
+            frac = 1.0 - (iteration - 1.0) / args.num_iterations
+            lrnow = frac * args.learning_rate
+            optimizer.param_groups[0]["lr"] = lrnow
+
+        # accumulate per-iteration episode successes
+        iter_successes = []
+        for step in range(0, args.num_steps):
+            global_step += args.num_envs
+            obs[step] = next_obs
+            dones[step] = next_done
+
+            with torch.no_grad():
+                action, logprob, _, value = agent.get_action_and_value(next_obs)
+                values[step] = value.flatten()
+            actions[step] = action
+            logprobs[step] = logprob
+
+            next_obs, reward, terminations, truncations, infos = envs.step(action.cpu().numpy())
+            next_done = np.logical_or(terminations, truncations)
+            rewards[step] = torch.tensor(reward).to(device).view(-1)
+            next_obs, next_done = torch.Tensor(next_obs).to(device), torch.Tensor(next_done).to(device)
+
+            # Episode stats logging similar to FrozenLake
+            try:
+                mask = None
+                if isinstance(infos, dict):
+                    if "_episode" in infos:
+                        mask = np.asarray(infos["_episode"]).astype(bool)
+                    elif "episode" in infos and isinstance(infos["episode"], dict) and "_l" in infos["episode"]:
+                        mask = np.asarray(infos["episode"]["_l"]).astype(bool)
+                if mask is not None and np.any(mask):
+                    r_arr = np.asarray(infos.get("episode", {}).get("r", np.zeros_like(mask, dtype=float)))
+                    l_arr = np.asarray(infos.get("episode", {}).get("l", np.zeros_like(mask, dtype=int)))
+                    # prefer success from info; fallback to ep return > 0
+                    if "success" in infos:
+                        succ_arr = np.asarray(infos.get("success", np.zeros_like(mask, dtype=bool))).astype(float)
+                    else:
+                        try:
+                            succ_arr = (np.asarray(r_arr) > 0).astype(float)
+                        except Exception:
+                            succ_arr = np.zeros_like(mask, dtype=float)
+                    # collect iteration successes for training success rate
+                    try:
+                        for s in np.asarray(succ_arr)[mask]:
+                            iter_successes.append(float(s))
+                    except Exception:
+                        pass
+                    if args.track:
+                        try:
+                            import wandb
+                            log_dict = {
+                                "global_step": int(global_step),
+                                "rollout/ep_rew_mean": float(np.mean(r_arr[mask])) if np.any(mask) else None,
+                                "rollout/ep_len_mean": float(np.mean(l_arr[mask])) if np.any(mask) else None,
+                                "rollout/success_rate": float(np.mean(succ_arr[mask])) if np.any(mask) else None,
+                            }
+                            wandb.log(log_dict, step=global_step)
+                        except Exception:
+                            pass
+            except Exception:
+                pass
+
+        # GAE
+        with torch.no_grad():
+            next_value = agent.get_value(next_obs).reshape(1, -1)
+            advantages = torch.zeros_like(rewards).to(device)
+            lastgaelam = 0
+            for t in reversed(range(args.num_steps)):
+                if t == args.num_steps - 1:
+                    nextnonterminal = 1.0 - next_done
+                    nextvalues = next_value
+                else:
+                    nextnonterminal = 1.0 - dones[t + 1]
+                    nextvalues = values[t + 1]
+                delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
+                advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
+            returns = advantages + values
+
+        # flatten batch
+        b_obs = obs.reshape((-1,) + envs.single_observation_space.shape)
+        b_logprobs = logprobs.reshape(-1)
+        b_actions = actions.reshape((-1,) + envs.single_action_space.shape)
+        b_advantages = advantages.reshape(-1)
+        b_returns = returns.reshape(-1)
+        b_values = values.reshape(-1)
+
+        # update
+        b_inds = np.arange(args.batch_size)
+        clipfracs = []
+        for epoch in range(args.update_epochs):
+            np.random.shuffle(b_inds)
+            for start in range(0, args.batch_size, args.minibatch_size):
+                end = start + args.minibatch_size
+                mb_inds = b_inds[start:end]
+
+                _, newlogprob, entropy, newvalue = agent.get_action_and_value(b_obs[mb_inds], b_actions.long()[mb_inds])
+                logratio = newlogprob - b_logprobs[mb_inds]
+                ratio = logratio.exp()
+
+                with torch.no_grad():
+                    old_approx_kl = (-logratio).mean()
+                    approx_kl = ((ratio - 1) - logratio).mean()
+                    clipfracs += [((ratio - 1.0).abs() > args.clip_coef).float().mean().item()]
+
+                mb_advantages = b_advantages[mb_inds]
+                if args.norm_adv:
+                    mb_advantages = (mb_advantages - mb_advantages.mean()) / (mb_advantages.std() + 1e-8)
+
+                pg_loss1 = -mb_advantages * ratio
+                pg_loss2 = -mb_advantages * torch.clamp(ratio, 1 - args.clip_coef, 1 + args.clip_coef)
+                pg_loss = torch.max(pg_loss1, pg_loss2).mean()
+
+                newvalue = newvalue.view(-1)
+                if args.clip_vloss:
+                    v_loss_unclipped = (newvalue - b_returns[mb_inds]) ** 2
+                    v_clipped = b_values[mb_inds] + torch.clamp(
+                        newvalue - b_values[mb_inds], -args.clip_coef, args.clip_coef,
+                    )
+                    v_loss_clipped = (v_clipped - b_returns[mb_inds]) ** 2
+                    v_loss = 0.5 * torch.max(v_loss_unclipped, v_loss_clipped).mean()
+                else:
+                    v_loss = 0.5 * ((newvalue - b_returns[mb_inds]) ** 2).mean()
+
+                entropy_loss = entropy.mean()
+                loss = pg_loss - args.ent_coef * entropy_loss + v_loss * args.vf_coef
+
+                optimizer.zero_grad()
+                loss.backward()
+                nn.utils.clip_grad_norm_(agent.parameters(), args.max_grad_norm)
+                optimizer.step()
+
+            if args.target_kl is not None and approx_kl > args.target_kl:
+                break
+
+        # metrics
+        y_pred, y_true = b_values.cpu().numpy(), b_returns.cpu().numpy()
+        var_y = np.var(y_true)
+        explained_var = np.nan if var_y == 0 else 1 - np.var(y_true - y_pred) / var_y
+
+        sps = int(global_step / (time.time() - start_time))
+        train_success_rate = float(np.mean(iter_successes)) if len(iter_successes) else 0.0
+        print(f"Iter {iteration:4d}/{args.num_iterations} | SPS: {sps:5d} | R: {rewards.mean().item():6.3f}")
+        if args.track:
+            try:
+                import wandb
+                wandb.log({
+                    "global_step": int(global_step),
+                    "charts/progress": float(100.0 * iteration / max(1, args.num_iterations)),
+                    "train/value_loss": float(v_loss.item()),
+                    "train/policy_loss": float(pg_loss.item()),
+                    "losses/value_loss": float(v_loss.item()),
+                    "losses/policy_loss": float(pg_loss.item()),
+                    "train/entropy": float(entropy_loss.item()),
+                    "train/old_approx_kl": float(old_approx_kl.item()),
+                    "train/approx_kl": float(approx_kl.item()),
+                    "train/clipfrac": float(np.mean(clipfracs)) if len(clipfracs) else 0.0,
+                    "losses/explained_variance": float(explained_var),
+                    "charts/avg_reward": float(rewards.mean().item()),
+                    "charts/avg_value": float(values.mean().item()),
+                    "perf/SPS": int(sps),
+                    "charts/SPS": int(sps),
+                    "train/success_rate": train_success_rate,
+                    "charts/train_success_rate": train_success_rate,
+                    "train/learning_rate": float(optimizer.param_groups[0]["lr"]),
+                }, step=global_step)
+            except Exception:
+                pass
+
+        # periodic evaluation collection
+        if iteration % eval_every_iters == 0:
+            try:
+                eval_thunk = make_env(0, run_name, args.seed + 9999, args.scramble_depth, args.max_steps_env, False)
+                collect_eval_trajectories(agent, eval_thunk, n_episodes=args.eval_episodes, step_tag=global_step)
+                if args.track:
+                    try:
+                        import json as _json
+                        from pathlib import Path as _Path
+                        mpath = _Path(f"runs/{run_name}/trajectories/step_{global_step}/metrics.json")
+                        if mpath.exists():
+                            with mpath.open("r") as mf:
+                                metrics = _json.load(mf)
+                            wandb.log({
+                                "eval/success_rate": metrics.get("success_rate"),
+                                "eval/avg_return": metrics.get("avg_return"),
+                                "eval/std_return": metrics.get("std_return"),
+                                "eval/episodes": metrics.get("episodes"),
+                            }, step=global_step)
+                    except Exception:
+                        pass
+                print(f"Collected {args.eval_episodes} eval trajectories at global_step {global_step}")
+            except Exception as e:
+                print(f"Warning: eval trajectory collection failed at step {global_step}: {e}")
+
+    envs.close()

wandb/run-20251218_185248-ts36d12x/files/config.yaml

@@ -0,0 +1,201 @@
+_wandb:
+    value:
+        cli_version: 0.23.0
+        code_path: code/cleanrl/cleanrl/ppo_rubikscube.py
+        e:
+            pa2plkmx4kakxqj0z9828uzvhc7cjmd1:
+                args:
+                    - --track
+                codePath: cleanrl/cleanrl/ppo_rubikscube.py
+                codePathLocal: cleanrl/cleanrl/ppo_rubikscube.py
+                cpu_count: 64
+                cpu_count_logical: 128
+                cudaVersion: "12.4"
+                disk:
+                    /:
+                        total: "5153960755200"
+                        used: "447561207808"
+                email: haoyu-wa22@mails.tsinghua.edu.cn
+                executable: /root/local/miniconda3/envs/ragen/bin/python
+                git:
+                    commit: 8d73639b99b38265453f898b8d6af7d4af50d56e
+                    remote: https://github.com/mll-lab-nu/RAGEN.git
+                gpu: NVIDIA H100 80GB HBM3
+                gpu_count: 8
+                gpu_nvidia:
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-35e2d43d-4067-82ce-90d4-def9e389bf28
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-af4135e3-88f2-e9ac-518d-502c75a85429
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-d7fdeeba-fe9b-ec03-d9f7-6724fe4266b5
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-ccc4f668-3882-5a8e-2c07-c5cd08f6f666
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-7b73c0cf-d3d5-e10c-7176-a43be1e41001
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-81b58d94-5d1f-8ec2-f9d2-fd56172ed177
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-03e8bc66-3b44-6794-49fd-5392fbdda6d1
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-b7cf0ec6-7c29-1179-dceb-09565da51890
+                host: pt-4d654cf4576f4d23ad3d3919f12932fe-worker-0
+                memory:
+                    total: "2163642122240"
+                os: Linux-5.14.0-284.25.1.el9_2.x86_64-x86_64-with-glibc2.35
+                program: /mnt/general/wanghy/RAGEN/cleanrl/cleanrl/ppo_rubikscube.py
+                python: CPython 3.10.19
+                root: /mnt/general/wanghy/RAGEN
+                startedAt: "2025-12-18T10:52:48.969828Z"
+                writerId: pa2plkmx4kakxqj0z9828uzvhc7cjmd1
+        m:
+            - "1": global_step
+              "6":
+                - 3
+              "7": []
+            - "2": charts/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": perf/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": train/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": rollout/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": eval/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+            - "2": losses/*
+              "5": 1
+              "6":
+                - 1
+              "7": []
+        python_version: 3.10.19
+        t:
+            "1":
+                - 1
+                - 11
+                - 30
+                - 49
+                - 50
+                - 51
+                - 105
+            "2":
+                - 1
+                - 11
+                - 30
+                - 49
+                - 50
+                - 51
+                - 105
+            "3":
+                - 7
+                - 13
+                - 16
+                - 61
+            "4": 3.10.19
+            "5": 0.23.0
+            "6": 4.57.1
+            "12": 0.23.0
+            "13": linux-x86_64
+anneal_lr:
+    value: true
+batch_size:
+    value: 1024
+capture_video:
+    value: false
+clip_coef:
+    value: 0.2
+clip_vloss:
+    value: true
+cuda:
+    value: true
+ent_coef:
+    value: 0.01
+env_id:
+    value: RubiksCube2x2
+eval_episodes:
+    value: 4000
+eval_splits:
+    value: 2
+exp_name:
+    value: ppo_rubikscube
+gae_lambda:
+    value: 0.95
+gamma:
+    value: 0.99
+learning_rate:
+    value: 0.00025
+max_grad_norm:
+    value: 0.5
+max_steps_env:
+    value: 20
+minibatch_size:
+    value: 256
+norm_adv:
+    value: true
+num_envs:
+    value: 8
+num_iterations:
+    value: 976
+num_minibatches:
+    value: 4
+num_steps:
+    value: 128
+scramble_depth:
+    value: 1
+seed:
+    value: 1
+target_kl:
+    value: null
+torch_deterministic:
+    value: true
+total_timesteps:
+    value: 1000000
+track:
+    value: true
+update_epochs:
+    value: 4
+vf_coef:
+    value: 0.5
+wandb_entity:
+    value: null
+wandb_project_name:
+    value: cleanRL

wandb/run-20251218_185248-ts36d12x/files/diff.patch

@@ -0,0 +1,1162 @@
+diff --git a/config/base.yaml b/config/base.yaml
+index a1659e4..9df6250 100644
+--- a/config/base.yaml
++++ b/config/base.yaml
+@@ -9,15 +9,16 @@ seed:
+   train: 10000
+   val: 123
+ 
+-micro_batch_size_per_gpu: 4
++micro_batch_size_per_gpu: 1
+ ppo_mini_batch_size: 32
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/Qwen/Qwen2.5-0.5B-Instruct
++# Qwen/Qwen2.5-0.5B-Instruct
+ enable_response_mask: True # Enabling response mask could improve stability of rollout/old_log_prob, as P(st|history) are no longer calculated in loss here. See https://docs.google.com/document/d/1bg7obeiKTExuHHBl5uOiSpec5uLDZ2Tgvxy6li5pHX4/edit?usp=sharing for more details.
+ grpo_advantage_length_weight: False # if you do not enable this and critic/advantage_estimator is GRPO, and the critic/advantages/mean is too low, then you can try enabling this to encourage reasoning and forbid collapse
+ 
+ lora:
+   rank: 0
+-  alpha: 64
++  alpha: 16
+   target_modules: all-linear
+ 
+ actor_rollout_ref:
+@@ -46,10 +47,10 @@ actor_rollout_ref:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu} # following micro_batch_size_per_gpu
+     tensor_model_parallel_size: 1
+-    max_model_len: 3600
++    max_model_len: 7200 #3600 why** 14400
+     prompt_length: 1 # useless. Just put it here
+     response_length: 400 # single-turn response length
+-    gpu_memory_utilization: 0.5
++    gpu_memory_utilization: 0.7
+     max_num_batched_tokens: 8192 # set only when enable_chunked_prefill is true
+     temperature: 1
+     rollout_filter_ratio: 0.25
+@@ -90,26 +91,28 @@ algorithm:
+     kl_coef: 0.000
+ 
+ trainer:
+-  project_name: ragen_latest
++  project_name: 
+   experiment_name: test
+   local_log_dir: "results/"
+-  save_freq: 100
++  save_freq: -1
+   total_training_steps: 200
+   validation_steps: 1 # validation instances = validation_steps * val_env_groups * group_size
+   val_before_train: True
+   n_gpus_per_node: 1
+   test_freq: 10
+   generations_to_log_to_wandb: 
++    train: 128
+     val: 20
+   logger: [ 'console', 'wandb' ]
+   max_actor_ckpt_to_keep: 1
+   max_critic_ckpt_to_keep: 1
++  default_local_dir: /mnt/general/wanghy/RAGEN/saves/
+ 
+ agent_proxy:
+   max_context_window: -1 # set a value > 0 to enable context window for long trajectory
+-  max_turn: 5
++  max_turn: 25 #25 why** 700
+   action_sep: "||"
+-  max_actions_per_turn: 2 # how many actions can be output at most in a single turn
++  max_actions_per_turn: 1 # how many actions can be output at most in a single turn
+   use_turn_scores: False # important to GAE when applying token-level rewards to token-level advantages. If False, will take the sum of scores as the reward for the last turn.
+   enable_think: True # False -> no think RL
+   reward_normalization:
+@@ -126,11 +129,11 @@ es_manager:
+       tags: ["CoordSokoban"]
+       n_groups: [8] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+   val:
+-    env_groups: 32
+-    group_size: 16 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
++    env_groups: 256
++    group_size: 1 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
+     env_configs:
+       tags: ["CoordSokoban"]
+-      n_groups: [32] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      n_groups: [256] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ ctx_manager:
+   generation: # go to vllm
+diff --git a/config/envs.yaml b/config/envs.yaml
+index 1002ac8..ba3ac6f 100644
+--- a/config/envs.yaml
++++ b/config/envs.yaml
+@@ -2,6 +2,7 @@ custom_envs:
+   SimpleSokoban:
+     env_type: sokoban
+     max_actions_per_traj: 10 # used in environment state manager to control the actual max actions executed per trajectory
++    # ORIGNAL env_instruction
+     env_instruction: |
+       You are solving the Sokoban puzzle. 
+       You are the player and you need to push all boxes to targets. 
+@@ -12,7 +13,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+ 
+   LargerSokoban:
+@@ -65,7 +66,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+       observation_format: "grid_coord"
+ 
+@@ -111,14 +112,14 @@ custom_envs:
+ 
+   FrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: "You are solving the FrozenLake puzzle. Forbid the whole and go to the target. You may move to the unintended direction due to the slippery ice. Example answer format: <think>To forbid the hole and go to the target, I should go left then go up.</think><answer>Left || Up</answer>"
+     max_tokens: 100
+     env_config: null
+ 
+   CoordFrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: |
+       You are solving the FrozenLake puzzle. The observation includes both a symbol grid and zero-indexed coordinates for the start, goal, player, and any holes.
+       Coordinates range from the top-left corner (0, 0) to the bottom-right corner (5, 5).
+@@ -195,4 +196,80 @@ custom_envs:
+     max_actions_per_traj: 30
+     env_instruction: "You are a Lean theorem prover. Given a Lean theorem statement, propose a sequence of tactics that completes the proof. Think step by step about which tactics to apply next. Provide tactics separated by '||', for example <answer>intro || simp || rfl</answer>."
+     max_tokens: 512
+-    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
+
++    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
++
++  
++  game_2048:
++    env_type: game_2048
++    max_actions_per_traj: 700
++    env_instruction: |
++      You are playing the 2048 game on a 4x4 grid. Merge equal tiles by sliding Up, Right, Down, or Left.
++      If a move is invalid (no tiles move), a small penalty is applied. Respond with a single action.
++      Example: <answer>Up</answer>
++    max_tokens: 8192
++    env_config: null
++
++  blackjack:
++    env_type: blackjack
++    max_actions_per_traj: 10
++    env_instruction: |
++      You are playing Blackjack against a dealer. The dealer must hit on 16 or less and stand on 17 or more.
++      Choose either Stick or Hit. Respond with a single action.
++      Example: <answer>Hit</answer>
++    max_tokens: 64
++    env_config: null
++
++  rubikscube:
++    env_type: rubikscube
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.
++      Available actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.
++      - Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).
++      - Modifiers: A letter alone means 90° clockwise (e.g., 'R'). A letter with prime (') means 90° counter-clockwise (e.g., "R'").
++      Respond with a sequence of actions separated by "||".
++      Example: <answer>U</answer>
++    max_tokens: 96
++    env_config:
++      scramble_depth: 1
++      max_steps: 20
++      render_mode: "text"
++  
++  SimpleSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 2x2 box contains the numbers 1-4 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format, for example: <answer>place 1 at row 2 col 3</answer> or <answer>1,2,3</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 4
++      difficulty: "easy"
++      render_format: "with_feedback"
++      show_conflicts: false
++      show_valid_numbers: false
++      max_steps: 20
++
++  MediumSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 30
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 3x3 box contains the numbers 1-9 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format: <answer>place 5 at row 2 col 3</answer> or <answer>2,3,5</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 9
++      difficulty: "medium"
++      render_format: "with_feedback"
++      show_conflicts: true
++      show_valid_numbers: true
++      max_steps: 81
++
+diff --git a/config/eval.yaml b/config/eval.yaml
+index e65f3a5..7775fd1 100644
+--- a/config/eval.yaml
++++ b/config/eval.yaml
+@@ -1,14 +1,15 @@
+ defaults:
+   - envs
+-
++enable_response_mask: True
+ system:
+-  CUDA_VISIBLE_DEVICES: "0"
++  CUDA_VISIBLE_DEVICES: "0,1,2,3,4,5,6,7"
+ 
+ seed:
+   train: 10000
+   val: 123
+ 
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/openai/gpt-oss-20b
++# /mnt/general/share/model/tyzhu/SPA-frozenlake-qwen2.5-1.5b-instruct
+ 
+ lora:
+   rank: 0
+@@ -24,9 +25,9 @@ actor_rollout_ref:
+   rollout:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu}
+-    tensor_model_parallel_size: 1
++    tensor_model_parallel_size: 8
+     dtype: bfloat16
+-    max_model_len: 3600
++    max_model_len: 7200
+     prompt_length: 1
+     response_length: 400
+     gpu_memory_utilization: 0.9
+@@ -35,6 +36,7 @@ actor_rollout_ref:
+     free_cache_engine: True
+     enable_chunked_prefill: False
+     disable_log_stats: False
++    do_sample: True
+     val_kwargs:
+       do_sample: True
+       temperature: 0.5
+@@ -44,9 +46,9 @@ actor_rollout_ref:
+ 
+ agent_proxy:
+   max_context_window: -1
+-  max_turn: 5
++  max_turn: 25
+   action_sep: "||"
+-  max_actions_per_turn: 2
++  max_actions_per_turn: 1
+   use_turn_scores: False
+   enable_think: True
+   reward_normalization:
+@@ -59,13 +61,13 @@ es_manager:
+     env_groups: 8
+     group_size: 16
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [8]
+   val:
+     env_groups: 32
+-    group_size: 16
++    group_size: 128
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [32]
+ 
+ ctx_manager:
+diff --git a/config/evaluate_api_llm.yaml b/config/evaluate_api_llm.yaml
+index b21c95e..e2e6761 100644
+--- a/config/evaluate_api_llm.yaml
++++ b/config/evaluate_api_llm.yaml
+@@ -1,8 +1,11 @@
++#export OPENAI_BASE_URL="https://api.ohmygpt.com/v1"
++#export OPENAI_API_KEY="sk-o4sMxBkN5BB100C4D4a3T3BlBkFJF7791CA39EA14ca98041"
++#python -m ragen.eval_api hydra.searchpath='[file://./verl/verl/trainer/config]'
+ defaults:
+   - base # this is a symbolic link to the verl/verl/trainer/config/ppo_trainer.yaml file
+ 
+ model_config:
+-  model_name: gpt-4o # should be registered in model_info
++  model_name: ark-deepseek-v3-250324 # should be registered in model_info
+   max_concurrency: 16
+ 
+ model_info:
+@@ -24,9 +27,9 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+-  gpt-4o:
++  gpt-4o-mini:
+     provider_name: openai
+-    model_name: gpt-4o
++    model_name: gpt-4o-mini
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+@@ -36,21 +39,42 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  ark-deepseek-v3-250324:
++    provider_name: openai
++    model_name: ark-deepseek-v3-250324
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
+   deepseek-v3:
+     provider_name: deepseek
+     model_name: deepseek-chat
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  glm-4.6:
++    provider_name: openai
++    model_name: glm-4.6
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
++  TA/openai/gpt-oss-120b:
++    provider_name: openai
++    model_name: TA/openai/gpt-oss-120b
++    generation_kwargs:
++      temperature: 0
++      max_tokens: 8192
++      # max_retries: 5
+ 
+-
+-
++agent_proxy:
++  max_turn: 5
+ es_manager:
+   val:
+-    env_groups: 256
++    env_groups: 128
+     group_size: 1 # should be set to 1 because val temperature is set to 0 and same prompt leads to same output
+     env_configs:
+-      tags: ["CoordSokoban"]
+-      n_groups: [256] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      tags: ["rubikscube"]
++      n_groups: [128] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ 
++rollout:
++  max_model_len: 7200
+
+Submodule external/webshop-minimal contains modified content
+diff --git a/external/webshop-minimal/requirements.txt b/external/webshop-minimal/requirements.txt
+index 5a1b04f..238ed5a 100644
+--- a/external/webshop-minimal/requirements.txt
++++ b/external/webshop-minimal/requirements.txt
+@@ -4,7 +4,7 @@ flask
+ html2text
+ rank_bm25
+ pyserini
+-faiss-cpu
++faiss-gpu
+ thefuzz
+ gdown
+ spacy
+diff --git a/ragen/env/__init__.py b/ragen/env/__init__.py
+index b0f3461..bc6838e 100644
+--- a/ragen/env/__init__.py
++++ b/ragen/env/__init__.py
+@@ -12,6 +12,14 @@ from .metamathqa.env import MetaMathQAEnv
+ from .metamathqa.config import MetaMathQAEnvConfig
+ from .lean.config import LeanEnvConfig
+ from .lean.env import LeanEnv
++from .game_2048.config import Game2048EnvConfig
++from .game_2048.env import Game2048Env
++from .blackjack.config import BlackjackEnvConfig
++from .blackjack.env import BlackjackEnv
++from .rubikscube.config import RubiksCube2x2Config
++from .rubikscube.env import RubiksCube2x2Env
++from .sudoku.config import SudokuEnvConfig
++from .sudoku.env import SudokuEnv
+ 
+ 
+ REGISTERED_ENVS = {
+@@ -22,6 +30,10 @@ REGISTERED_ENVS = {
+     # 'alfworld': AlfredTXTEnv,
+     'metamathqa': MetaMathQAEnv,
+     'lean': LeanEnv,
++    'game_2048': Game2048Env,
++    'blackjack': BlackjackEnv,
++    'rubikscube': RubiksCube2x2Env,
++    'sudoku': SudokuEnv,
+ }
+ 
+ REGISTERED_ENV_CONFIGS = {
+@@ -32,6 +44,10 @@ REGISTERED_ENV_CONFIGS = {
+     # 'alfworld': AlfredEnvConfig,
+     'metamathqa': MetaMathQAEnvConfig,
+     'lean': LeanEnvConfig,
++    'game_2048': Game2048EnvConfig,
++    'blackjack': BlackjackEnvConfig,
++    'rubikscube': RubiksCube2x2Config,
++    'sudoku': SudokuEnvConfig,
+ }
+ 
+ try:
+diff --git a/ragen/env/frozen_lake/config.py b/ragen/env/frozen_lake/config.py
+index de054f4..9950c34 100644
+--- a/ragen/env/frozen_lake/config.py
++++ b/ragen/env/frozen_lake/config.py
+@@ -8,7 +8,7 @@ class FrozenLakeEnvConfig:
+     size: int = 4
+     p: float = 0.9
+     success_rate: float = 0.8
+-    is_slippery: bool = True
++    is_slippery: bool = False
+     map_seed: Optional[int] = None
+     render_mode: str = "text"
+     observation_format: str = "grid"
+diff --git a/ragen/env/frozen_lake/env.py b/ragen/env/frozen_lake/env.py
+index 9ee3add..eef11d8 100644
+--- a/ragen/env/frozen_lake/env.py
++++ b/ragen/env/frozen_lake/env.py
+@@ -13,6 +13,7 @@ from ragen.env.base import BaseDiscreteActionEnv
+ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+     def __init__(self, config: FrozenLakeEnvConfig = FrozenLakeEnvConfig()):
+         # Using mappings directly from config
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.GRID_LOOKUP = config.grid_lookup
+         self.ACTION_LOOKUP = config.action_lookup
+@@ -95,6 +96,7 @@ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+ if __name__ == "__main__":
+     import matplotlib.pyplot as plt
+     config = FrozenLakeEnvConfig(size=4, is_slippery=True)
++    # import pdb;pdb.set_trace()
+     env = FrozenLakeEnv(config)
+     print(env.reset())
+     while True:
+diff --git a/ragen/env/sokoban/env.py b/ragen/env/sokoban/env.py
+index 17cd636..6922b54 100644
+--- a/ragen/env/sokoban/env.py
++++ b/ragen/env/sokoban/env.py
+@@ -33,6 +33,7 @@ class SokobanEnv(BaseDiscreteActionEnv, GymSokobanEnv):
+     def reset(self, seed=None, mode=None):
+         try:
+             with all_seed(seed):
++                # import pdb;pdb.set_trace()
+                 self.room_fixed, self.room_state, self.box_mapping, action_sequence = generate_room(
+                     dim=self.dim_room,
+                     num_steps=self.num_gen_steps,
+diff --git a/ragen/llm_agent/agent_proxy.py b/ragen/llm_agent/agent_proxy.py
+index c25b50d..f6c4088 100644
+--- a/ragen/llm_agent/agent_proxy.py
++++ b/ragen/llm_agent/agent_proxy.py
+@@ -287,6 +287,7 @@ def main(config):
+ 	proxy = LLMAgentProxy(config, actor_wg, tokenizer)
+ 	import time
+ 	start_time = time.time()
++	# import pdb;pdb.set_trace()
+ 	rollouts = proxy.rollout(
+ 		DataProto(
+ 			batch=None,
+diff --git a/ragen/llm_agent/base_llm.py b/ragen/llm_agent/base_llm.py
+index 358a2eb..b6f7797 100644
+--- a/ragen/llm_agent/base_llm.py
++++ b/ragen/llm_agent/base_llm.py
+@@ -33,6 +33,7 @@ class OpenAIProvider(LLMProvider):
+             raise ValueError("OpenAI API key not provided and not found in environment variables")
+         
+         self.client = AsyncOpenAI(api_key=self.api_key)
++        # import pdb;pdb.set_trace()
+     
+     async def generate(self, messages: List[Dict[str, str]], **kwargs) -> LLMResponse:
+         if "o1-mini" in self.model_name:
+@@ -46,10 +47,7 @@ class OpenAIProvider(LLMProvider):
+         )
+         if response.choices[0].finish_reason in ['length', 'content_filter']:
+             raise ValueError("Content filtered or length exceeded")
+-        return LLMResponse(
+-            content=response.choices[0].message.content,
+-            model_name=response.model
+-        )
++        return LLMResponse(content=response.choices[0].message.content,model_name=response.model)
+ 
+ class DeepSeekProvider(LLMProvider):
+     """DeepSeek API provider implementation"""
+@@ -199,7 +197,7 @@ class ConcurrentLLM:
+         
+         # Queue to store unfinished or failed tasks
+         current_batch = messages_list.copy()
+-        max_retries = kwargs.get("max_retries", 100)
++        max_retries = kwargs.get("max_retries", 10)
+         retry_count = 0
+         
+         while current_batch and retry_count < max_retries:
+diff --git a/ragen/llm_agent/ctx_manager.py b/ragen/llm_agent/ctx_manager.py
+index 905247a..af20a01 100644
+--- a/ragen/llm_agent/ctx_manager.py
++++ b/ragen/llm_agent/ctx_manager.py
+@@ -88,6 +88,7 @@ class ContextManager:
+         Initialize the ContextManager.
+         Processor is used to process the image data.
+         """
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.tokenizer = tokenizer
+         self.processor = processor
+@@ -311,7 +312,7 @@ class ContextManager:
+ 
+             llm_input_texts.append(text_with_prompt)
+             messages_list.append(messages)
+-
++        # import pdb;pdb.set_trace()
+         inputs = self.tokenizer(llm_input_texts, return_tensors="pt", padding=True, padding_side="left", truncation=False) # We have truncated previously, truncation in tokenizer may cause issues.
+         input_ids, attention_mask = inputs.input_ids, inputs.attention_mask
+         position_ids = (attention_mask.cumsum(dim=-1) - 1).clamp(min=0)
+@@ -353,6 +354,30 @@ class ContextManager:
+                 key: np.sum(value) / self.env_nums[key.split("/")[0]]
+                 for key, value in metrics.items()
+             }
++            # Derived metrics for wandb logging
++            try:
++                # charts/avg_episode_return: average across all 2048 env groups
++                ep_keys = [k for k in metrics.keys() if k.endswith('/episodic_return')]
++                if len(ep_keys) > 0:
++                    ep_vals = []
++                    for k in ep_keys:
++                        tag = k.split('/')[0]
++                        denom = self.env_nums.get(tag, max(1, len(metrics[k])))
++                        ep_vals.append(float(np.sum(metrics[k]) / denom))
++                    mean_metrics["charts/avg_episode_return"] = float(np.mean(ep_vals))
++            except Exception:
++                pass
++            try:
++                # rollout/max_tile: max over all envs in this batch
++                tile_keys = [k for k in metrics.keys() if k.endswith('/max_tile')]
++                if len(tile_keys) > 0:
++                    tile_vals = []
++                    for k in tile_keys:
++                        tile_vals.extend(metrics[k])
++                    if len(tile_vals) > 0:
++                        mean_metrics["rollout/max_tile"] = int(np.max(tile_vals))
++            except Exception:
++                pass
+             for key, values in metrics.items():
+                 if not isinstance(values, list):
+                     continue
+diff --git a/ragen/llm_agent/es_manager.py b/ragen/llm_agent/es_manager.py
+index b87a2b3..1a49d3e 100644
+--- a/ragen/llm_agent/es_manager.py
++++ b/ragen/llm_agent/es_manager.py
+@@ -128,18 +128,40 @@ class EnvStateManager:
+         env_outputs: List[Dict]
+             {env_id: int, history: List[Dict][{state: str, actions: List[str], reward: float, info: Dict, llm_response: str, llm_raw_response: str, (Optional)images: List[PIL.Image.Image]}]}
+         """
++        # def _execute_actions(env, actions):
++        #     acc_reward, turn_info, turn_done = 0, {}, False
++        #     executed_actions = []
++        #     for action in actions:
++        #         _, reward, done, info = env.step(action)
++        #         acc_reward += reward
++        #         turn_info.update(info) # NOTE: currently use last info for multi-action
++        #         executed_actions.append(action)
++        #         if done:
++        #             turn_done = True
++        #             break
++            
++        #     return acc_reward, turn_info, turn_done, executed_actions
+         def _execute_actions(env, actions):
+-            acc_reward, turn_info, turn_done = 0, {}, False
++            acc_reward, turn_info, turn_done = 0.0, {}, False
++            raw_acc_reward = 0.0
+             executed_actions = []
+             for action in actions:
+                 _, reward, done, info = env.step(action)
+-                acc_reward += reward
++                acc_reward += float(reward)
++                try:
++                    raw_acc_reward += float(info.get('raw_reward', 0.0))
++                except Exception:
++                    pass
+                 turn_info.update(info) # NOTE: currently use last info for multi-action
+                 executed_actions.append(action)
+                 if done:
+                     turn_done = True
+                     break
+-            
++            # Overwrite per-turn raw_reward to reflect the sum across all executed actions in this turn
++            try:
++                turn_info['raw_reward'] = float(raw_acc_reward)
++            except Exception:
++                pass
+             return acc_reward, turn_info, turn_done, executed_actions
+ 
+         def _log_env_state(status, history, cur_obs, max_actions_per_traj, executed_actions, all_actions, acc_reward, turn_done, turn_info, env_input):
+@@ -198,6 +220,18 @@ class EnvStateManager:
+                 'success': float(status.terminated and (not status.truncated)),
+                 'num_actions': status.num_actions,
+             }
++            # Add episodic-level metrics
++            # try:
++            #     # Sum of per-turn rewards equals the episodic return (env-shaped reward)
++            #     env_metric['episodic_return'] = float(sum(status.rewards))
++            # except Exception:
++            #     pass
++            try:
++                # Final max tile on the board at the end of the rollout
++                import numpy as _np
++                env_metric['max_tile'] = int(_np.max(entry['env'].grid))
++            except Exception:
++                pass
+             custom_metric = {}
+             for turn in cache['history']:
+                 for k, v in turn.get('info', {}).items():
+@@ -212,6 +246,12 @@ class EnvStateManager:
+                             "Skipping non-numeric metric '%s' with value %r for env %s.",
+                             k, v, entry['tag']
+                         )
++            # Add episodic_return as the SUM of raw_reward across steps (align with CleanRL)
++            try:
++                if 'raw_reward' in custom_metric:
++                    env_metric['episodic_return'] = float(np.sum(custom_metric['raw_reward']))
++            except Exception:
++                pass
+             for k, v in custom_metric.items():
+                 # TODO: Move TURN_LVL_METRICS into the environment
+                 if "webshop" not in cache['tag'].lower() or ("webshop" in cache['tag'].lower() and k in TURN_LVL_METRICS):
+@@ -219,7 +259,12 @@ class EnvStateManager:
+                 else:
+                     env_metric['traj_sum/' + k] = np.sum(v)
+ 
+-
++            try:
++                if 'score' in custom_metric and len(custom_metric['score']) > 0:
++                    env_metric['final_score'] = float(custom_metric['score'][-1])
++            except Exception:
++                pass
++                
+             cache['history'][-1]['metrics'] = custom_metric
+             env_metric = {f"{entry['tag']}/{k}": v for k, v in env_metric.items()}
+             cache['metrics'] = env_metric
+diff --git a/requirements.txt b/requirements.txt
+index 2fd756e..9b521bf 100644
+--- a/requirements.txt
++++ b/requirements.txt
+@@ -7,7 +7,6 @@ accelerate
+ codetiming
+ datasets
+ dill
+-flash-attn==2.7.4.post1
+ hydra-core
+ numpy
+ pandas
+@@ -15,19 +14,19 @@ pybind11
+ ray>=2.10
+ tensordict>=0.8.0,<0.9.0
+ transformers
+-vllm==0.8.2
++vllm==0.8.5
+ wandb
+ gymnasium
+ gymnasium[toy-text]
+ 
+ pyarrow>=15.0.0
+ pylatexenc
+-torchdata
++# torchdata
+ debugpy
+ 
+ together
+ anthropic
+-faiss-cpu==1.11.0
++faiss-gpu
+ 
+ # This is optional, but needs to be installed with main requirements if you want to use webshop
+ -r external/webshop-minimal/requirements.txt
+diff --git a/scripts/setup_ragen.sh b/scripts/setup_ragen.sh
+index f9a7cd9..85a93c4 100644
+--- a/scripts/setup_ragen.sh
++++ b/scripts/setup_ragen.sh
+@@ -94,10 +94,10 @@ main() {
+         pip install torch==2.5.0 --index-url https://download.pytorch.org/whl/cu124
+         
+         print_step "Installing flash-attention..."
+-        pip3 install flash-attn==2.7.4.post1 --no-build-isolation
++        # pip3 install flash-attn==2.7.4.post1 --no-build-isolation
+     else
+         print_step "Installing PyTorch without CUDA support..."
+-        pip install torch==2.5.0
++        pip install torch==2.4.0
+     fi
+     
+     # Install remaining requirements
+@@ -137,8 +137,8 @@ main() {
+     conda install conda-forge::gdown
+     mkdir -p external/webshop-minimal/webshop_minimal/data/full
+     cd external/webshop-minimal/webshop_minimal/data/full
+-    gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
+-    gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
++    # gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
++    # gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
+     cd ../../../../..
+ 
+     echo -e "${GREEN}Installation completed successfully!${NC}"
+diff --git a/train_all.sh b/train_all.sh
+index 0157306..33035f5 100755
+--- a/train_all.sh
++++ b/train_all.sh
+@@ -6,246 +6,257 @@ USE_GRPO="algorithm.adv_estimator=grpo"
+ USE_PPO="algorithm.adv_estimator=gae" # by default.
+ USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+ 
++
++# python train.py --config-name _8_2048 system.CUDA_VISIBLE_DEVICES="'0,1,2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=game_2048 $USE_PPO $USE_BASE 
++
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-1 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-1.5B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'4,5'" trainer.project_name=ragen_latest_qwen_25_3b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-3B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-3 $USE_PPO $USE_BASE 
++
+ # Section 3.1&3.2 - General Observations
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozen_lake-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozen_lake-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=bandit-ppo-multitask $USE_PPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'7'" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-ppo-frommlp $USE_PPO $USE_BASE 
+ 
+-# Section 4.1 - Filtering and critic
+-# 0.25
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-
+-wait
+-
+-# 0.5
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-
+-# 0.75
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-
+-wait
+-
+-# Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=sokoban-ppo-box1-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-wait
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.project_name=ragen_latest_qwen_25_3b_it  trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_3B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-# Section 5.1 - Reasoning Helps Generalization
++# # Section 4.1 - Filtering and critic
++# # 0.25
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
+-    agent_proxy.enable_think=False
++# # 0.5
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++
++# # 0.75
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
+-    agent_proxy.enable_think=False
++# # Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+ 
++# wait
+ 
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
++# # Section 5.1 - Reasoning Helps Generalization
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
+ 
+-# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
++#     agent_proxy.enable_think=False
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader"
+ 
+-# COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
++#     agent_proxy.enable_think=False
+ 
+-wait
+ 
+-# Section 5.2 - what leads to better reasoning?
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
++# # SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
++# # COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
+ 
++# wait
+ 
+-wait
++# # Section 5.2 - what leads to better reasoning?
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
+ 
+-# Section 6.1 varying action count
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
++# wait
+ 
+-# section 6.2 Varying prompt diversity
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
+ 
+-wait 
++# wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
++# wait
+ 
++# # Section 6.1 varying action count
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
+ 
++# # section 6.2 Varying prompt diversity
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
+ 
++# wait 
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
+ 
+-# Extension: Training 7B reasoning model
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
+ 
+-# trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
++# # Extension: Training 7B reasoning model
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
+ 
++# python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
+ 
++# # trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_PPO="algorithm.adv_estimator=gae" # by default.
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
+-# enable_response_mask: False
+-# grpo_advantage_length_weight: True
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+ 
+-# extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
++# # USE_PPO="algorithm.adv_estimator=gae" # by default.
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
+ 
+-# extension: bi-level gae
+-python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
+-    system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
+-    model_path=Qwen/Qwen2.5-0.5B-Instruct \
+-    algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
+-    agent_proxy.use_turn_scores=True \
+-    actor_rollout_ref.rollout.tp_size_check=False
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
+ 
+-# extension: webshop
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # enable_response_mask: False
++# # grpo_advantage_length_weight: True
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: bi-level gae
++# python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
++#     system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
++#     model_path=Qwen/Qwen2.5-0.5B-Instruct \
++#     algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
++#     agent_proxy.use_turn_scores=True \
++#     actor_rollout_ref.rollout.tp_size_check=False
+ 
+ 
+-# normal:sokoban
+-# extension: sokoban
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # extension: webshop
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
+ 
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++
++# # normal:sokoban
++# # extension: sokoban
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
++
++
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
++
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+ 
+ 
+-python train.py \
+-    trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+
++# python train.py \
++#     trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+

wandb/run-20251218_185248-ts36d12x/files/diff_8d73639b99b38265453f898b8d6af7d4af50d56e.patch

@@ -0,0 +1,1162 @@
+diff --git a/config/base.yaml b/config/base.yaml
+index a1659e4..9df6250 100644
+--- a/config/base.yaml
++++ b/config/base.yaml
+@@ -9,15 +9,16 @@ seed:
+   train: 10000
+   val: 123
+ 
+-micro_batch_size_per_gpu: 4
++micro_batch_size_per_gpu: 1
+ ppo_mini_batch_size: 32
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/Qwen/Qwen2.5-0.5B-Instruct
++# Qwen/Qwen2.5-0.5B-Instruct
+ enable_response_mask: True # Enabling response mask could improve stability of rollout/old_log_prob, as P(st|history) are no longer calculated in loss here. See https://docs.google.com/document/d/1bg7obeiKTExuHHBl5uOiSpec5uLDZ2Tgvxy6li5pHX4/edit?usp=sharing for more details.
+ grpo_advantage_length_weight: False # if you do not enable this and critic/advantage_estimator is GRPO, and the critic/advantages/mean is too low, then you can try enabling this to encourage reasoning and forbid collapse
+ 
+ lora:
+   rank: 0
+-  alpha: 64
++  alpha: 16
+   target_modules: all-linear
+ 
+ actor_rollout_ref:
+@@ -46,10 +47,10 @@ actor_rollout_ref:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu} # following micro_batch_size_per_gpu
+     tensor_model_parallel_size: 1
+-    max_model_len: 3600
++    max_model_len: 7200 #3600 why** 14400
+     prompt_length: 1 # useless. Just put it here
+     response_length: 400 # single-turn response length
+-    gpu_memory_utilization: 0.5
++    gpu_memory_utilization: 0.7
+     max_num_batched_tokens: 8192 # set only when enable_chunked_prefill is true
+     temperature: 1
+     rollout_filter_ratio: 0.25
+@@ -90,26 +91,28 @@ algorithm:
+     kl_coef: 0.000
+ 
+ trainer:
+-  project_name: ragen_latest
++  project_name: 
+   experiment_name: test
+   local_log_dir: "results/"
+-  save_freq: 100
++  save_freq: -1
+   total_training_steps: 200
+   validation_steps: 1 # validation instances = validation_steps * val_env_groups * group_size
+   val_before_train: True
+   n_gpus_per_node: 1
+   test_freq: 10
+   generations_to_log_to_wandb: 
++    train: 128
+     val: 20
+   logger: [ 'console', 'wandb' ]
+   max_actor_ckpt_to_keep: 1
+   max_critic_ckpt_to_keep: 1
++  default_local_dir: /mnt/general/wanghy/RAGEN/saves/
+ 
+ agent_proxy:
+   max_context_window: -1 # set a value > 0 to enable context window for long trajectory
+-  max_turn: 5
++  max_turn: 25 #25 why** 700
+   action_sep: "||"
+-  max_actions_per_turn: 2 # how many actions can be output at most in a single turn
++  max_actions_per_turn: 1 # how many actions can be output at most in a single turn
+   use_turn_scores: False # important to GAE when applying token-level rewards to token-level advantages. If False, will take the sum of scores as the reward for the last turn.
+   enable_think: True # False -> no think RL
+   reward_normalization:
+@@ -126,11 +129,11 @@ es_manager:
+       tags: ["CoordSokoban"]
+       n_groups: [8] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+   val:
+-    env_groups: 32
+-    group_size: 16 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
++    env_groups: 256
++    group_size: 1 # should be set to 1 because when val temperature is set to 0 and group size > 1, there will be repetitive prompts which leads to same trajectory.
+     env_configs:
+       tags: ["CoordSokoban"]
+-      n_groups: [32] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      n_groups: [256] # TODO: If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ ctx_manager:
+   generation: # go to vllm
+diff --git a/config/envs.yaml b/config/envs.yaml
+index 1002ac8..ba3ac6f 100644
+--- a/config/envs.yaml
++++ b/config/envs.yaml
+@@ -2,6 +2,7 @@ custom_envs:
+   SimpleSokoban:
+     env_type: sokoban
+     max_actions_per_traj: 10 # used in environment state manager to control the actual max actions executed per trajectory
++    # ORIGNAL env_instruction
+     env_instruction: |
+       You are solving the Sokoban puzzle. 
+       You are the player and you need to push all boxes to targets. 
+@@ -12,7 +13,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+ 
+   LargerSokoban:
+@@ -65,7 +66,7 @@ custom_envs:
+     env_config: # keys should be a subset of SokobanConfig
+       dim_x: 6
+       dim_y: 6
+-      num_boxes: 1
++      num_boxes: 2
+       max_steps: 100
+       observation_format: "grid_coord"
+ 
+@@ -111,14 +112,14 @@ custom_envs:
+ 
+   FrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: "You are solving the FrozenLake puzzle. Forbid the whole and go to the target. You may move to the unintended direction due to the slippery ice. Example answer format: <think>To forbid the hole and go to the target, I should go left then go up.</think><answer>Left || Up</answer>"
+     max_tokens: 100
+     env_config: null
+ 
+   CoordFrozenLake:
+     env_type: frozen_lake
+-    max_actions_per_traj: 10
++    max_actions_per_traj: 25
+     env_instruction: |
+       You are solving the FrozenLake puzzle. The observation includes both a symbol grid and zero-indexed coordinates for the start, goal, player, and any holes.
+       Coordinates range from the top-left corner (0, 0) to the bottom-right corner (5, 5).
+@@ -195,4 +196,80 @@ custom_envs:
+     max_actions_per_traj: 30
+     env_instruction: "You are a Lean theorem prover. Given a Lean theorem statement, propose a sequence of tactics that completes the proof. Think step by step about which tactics to apply next. Provide tactics separated by '||', for example <answer>intro || simp || rfl</answer>."
+     max_tokens: 512
+-    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
+
++    env_config: null # Please refer to ragen/env/lean/config.py for a full list of parameters.
++
++  
++  game_2048:
++    env_type: game_2048
++    max_actions_per_traj: 700
++    env_instruction: |
++      You are playing the 2048 game on a 4x4 grid. Merge equal tiles by sliding Up, Right, Down, or Left.
++      If a move is invalid (no tiles move), a small penalty is applied. Respond with a single action.
++      Example: <answer>Up</answer>
++    max_tokens: 8192
++    env_config: null
++
++  blackjack:
++    env_type: blackjack
++    max_actions_per_traj: 10
++    env_instruction: |
++      You are playing Blackjack against a dealer. The dealer must hit on 16 or less and stand on 17 or more.
++      Choose either Stick or Hit. Respond with a single action.
++      Example: <answer>Hit</answer>
++    max_tokens: 64
++    env_config: null
++
++  rubikscube:
++    env_type: rubikscube
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.
++      Available actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.
++      - Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).
++      - Modifiers: A letter alone means 90° clockwise (e.g., 'R'). A letter with prime (') means 90° counter-clockwise (e.g., "R'").
++      Respond with a sequence of actions separated by "||".
++      Example: <answer>U</answer>
++    max_tokens: 96
++    env_config:
++      scramble_depth: 1
++      max_steps: 20
++      render_mode: "text"
++  
++  SimpleSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 20
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 2x2 box contains the numbers 1-4 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format, for example: <answer>place 1 at row 2 col 3</answer> or <answer>1,2,3</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 4
++      difficulty: "easy"
++      render_format: "with_feedback"
++      show_conflicts: false
++      show_valid_numbers: false
++      max_steps: 20
++
++  MediumSudoku:
++    env_type: sudoku
++    max_actions_per_traj: 30
++    env_instruction: |
++      You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 3x3 box contains the numbers 1-9 without repetition.
++      Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
++      Place numbers one at a time using the format: <answer>place 5 at row 2 col 3</answer> or <answer>2,3,5</answer>
++      The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
++    max_tokens: 150
++    parallel_friendly: false
++    max_workers: 32
++    env_config:
++      grid_size: 9
++      difficulty: "medium"
++      render_format: "with_feedback"
++      show_conflicts: true
++      show_valid_numbers: true
++      max_steps: 81
++
+diff --git a/config/eval.yaml b/config/eval.yaml
+index e65f3a5..7775fd1 100644
+--- a/config/eval.yaml
++++ b/config/eval.yaml
+@@ -1,14 +1,15 @@
+ defaults:
+   - envs
+-
++enable_response_mask: True
+ system:
+-  CUDA_VISIBLE_DEVICES: "0"
++  CUDA_VISIBLE_DEVICES: "0,1,2,3,4,5,6,7"
+ 
+ seed:
+   train: 10000
+   val: 123
+ 
+-model_path: Qwen/Qwen2.5-3B-Instruct
++model_path: /mnt/general/share/model/openai/gpt-oss-20b
++# /mnt/general/share/model/tyzhu/SPA-frozenlake-qwen2.5-1.5b-instruct
+ 
+ lora:
+   rank: 0
+@@ -24,9 +25,9 @@ actor_rollout_ref:
+   rollout:
+     name: vllm
+     log_prob_micro_batch_size_per_gpu: ${micro_batch_size_per_gpu}
+-    tensor_model_parallel_size: 1
++    tensor_model_parallel_size: 8
+     dtype: bfloat16
+-    max_model_len: 3600
++    max_model_len: 7200
+     prompt_length: 1
+     response_length: 400
+     gpu_memory_utilization: 0.9
+@@ -35,6 +36,7 @@ actor_rollout_ref:
+     free_cache_engine: True
+     enable_chunked_prefill: False
+     disable_log_stats: False
++    do_sample: True
+     val_kwargs:
+       do_sample: True
+       temperature: 0.5
+@@ -44,9 +46,9 @@ actor_rollout_ref:
+ 
+ agent_proxy:
+   max_context_window: -1
+-  max_turn: 5
++  max_turn: 25
+   action_sep: "||"
+-  max_actions_per_turn: 2
++  max_actions_per_turn: 1
+   use_turn_scores: False
+   enable_think: True
+   reward_normalization:
+@@ -59,13 +61,13 @@ es_manager:
+     env_groups: 8
+     group_size: 16
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [8]
+   val:
+     env_groups: 32
+-    group_size: 16
++    group_size: 128
+     env_configs:
+-      tags: ["CoordSokoban"]
++      tags: ["BanditTest"]
+       n_groups: [32]
+ 
+ ctx_manager:
+diff --git a/config/evaluate_api_llm.yaml b/config/evaluate_api_llm.yaml
+index b21c95e..e2e6761 100644
+--- a/config/evaluate_api_llm.yaml
++++ b/config/evaluate_api_llm.yaml
+@@ -1,8 +1,11 @@
++#export OPENAI_BASE_URL="https://api.ohmygpt.com/v1"
++#export OPENAI_API_KEY="sk-o4sMxBkN5BB100C4D4a3T3BlBkFJF7791CA39EA14ca98041"
++#python -m ragen.eval_api hydra.searchpath='[file://./verl/verl/trainer/config]'
+ defaults:
+   - base # this is a symbolic link to the verl/verl/trainer/config/ppo_trainer.yaml file
+ 
+ model_config:
+-  model_name: gpt-4o # should be registered in model_info
++  model_name: ark-deepseek-v3-250324 # should be registered in model_info
+   max_concurrency: 16
+ 
+ model_info:
+@@ -24,9 +27,9 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+-  gpt-4o:
++  gpt-4o-mini:
+     provider_name: openai
+-    model_name: gpt-4o
++    model_name: gpt-4o-mini
+     generation_kwargs:
+       temperature: 0
+       max_tokens: 512 # max_completion_tokens if o1-mini
+@@ -36,21 +39,42 @@ model_info:
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  ark-deepseek-v3-250324:
++    provider_name: openai
++    model_name: ark-deepseek-v3-250324
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
+   deepseek-v3:
+     provider_name: deepseek
+     model_name: deepseek-chat
+     generation_kwargs:
+       temperature: 0
+       max_completion_tokens: 512
++  glm-4.6:
++    provider_name: openai
++    model_name: glm-4.6
++    generation_kwargs:
++      temperature: 0
++      max_completion_tokens: 512
++  TA/openai/gpt-oss-120b:
++    provider_name: openai
++    model_name: TA/openai/gpt-oss-120b
++    generation_kwargs:
++      temperature: 0
++      max_tokens: 8192
++      # max_retries: 5
+ 
+-
+-
++agent_proxy:
++  max_turn: 5
+ es_manager:
+   val:
+-    env_groups: 256
++    env_groups: 128
+     group_size: 1 # should be set to 1 because val temperature is set to 0 and same prompt leads to same output
+     env_configs:
+-      tags: ["CoordSokoban"]
+-      n_groups: [256] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
++      tags: ["rubikscube"]
++      n_groups: [128] # If not set, all env names divide nums equally. Under the same group, the env config and env seed (prompt) are equal in each generation
+ 
+ 
++rollout:
++  max_model_len: 7200
+
+Submodule external/webshop-minimal contains modified content
+diff --git a/external/webshop-minimal/requirements.txt b/external/webshop-minimal/requirements.txt
+index 5a1b04f..238ed5a 100644
+--- a/external/webshop-minimal/requirements.txt
++++ b/external/webshop-minimal/requirements.txt
+@@ -4,7 +4,7 @@ flask
+ html2text
+ rank_bm25
+ pyserini
+-faiss-cpu
++faiss-gpu
+ thefuzz
+ gdown
+ spacy
+diff --git a/ragen/env/__init__.py b/ragen/env/__init__.py
+index b0f3461..bc6838e 100644
+--- a/ragen/env/__init__.py
++++ b/ragen/env/__init__.py
+@@ -12,6 +12,14 @@ from .metamathqa.env import MetaMathQAEnv
+ from .metamathqa.config import MetaMathQAEnvConfig
+ from .lean.config import LeanEnvConfig
+ from .lean.env import LeanEnv
++from .game_2048.config import Game2048EnvConfig
++from .game_2048.env import Game2048Env
++from .blackjack.config import BlackjackEnvConfig
++from .blackjack.env import BlackjackEnv
++from .rubikscube.config import RubiksCube2x2Config
++from .rubikscube.env import RubiksCube2x2Env
++from .sudoku.config import SudokuEnvConfig
++from .sudoku.env import SudokuEnv
+ 
+ 
+ REGISTERED_ENVS = {
+@@ -22,6 +30,10 @@ REGISTERED_ENVS = {
+     # 'alfworld': AlfredTXTEnv,
+     'metamathqa': MetaMathQAEnv,
+     'lean': LeanEnv,
++    'game_2048': Game2048Env,
++    'blackjack': BlackjackEnv,
++    'rubikscube': RubiksCube2x2Env,
++    'sudoku': SudokuEnv,
+ }
+ 
+ REGISTERED_ENV_CONFIGS = {
+@@ -32,6 +44,10 @@ REGISTERED_ENV_CONFIGS = {
+     # 'alfworld': AlfredEnvConfig,
+     'metamathqa': MetaMathQAEnvConfig,
+     'lean': LeanEnvConfig,
++    'game_2048': Game2048EnvConfig,
++    'blackjack': BlackjackEnvConfig,
++    'rubikscube': RubiksCube2x2Config,
++    'sudoku': SudokuEnvConfig,
+ }
+ 
+ try:
+diff --git a/ragen/env/frozen_lake/config.py b/ragen/env/frozen_lake/config.py
+index de054f4..9950c34 100644
+--- a/ragen/env/frozen_lake/config.py
++++ b/ragen/env/frozen_lake/config.py
+@@ -8,7 +8,7 @@ class FrozenLakeEnvConfig:
+     size: int = 4
+     p: float = 0.9
+     success_rate: float = 0.8
+-    is_slippery: bool = True
++    is_slippery: bool = False
+     map_seed: Optional[int] = None
+     render_mode: str = "text"
+     observation_format: str = "grid"
+diff --git a/ragen/env/frozen_lake/env.py b/ragen/env/frozen_lake/env.py
+index 9ee3add..eef11d8 100644
+--- a/ragen/env/frozen_lake/env.py
++++ b/ragen/env/frozen_lake/env.py
+@@ -13,6 +13,7 @@ from ragen.env.base import BaseDiscreteActionEnv
+ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+     def __init__(self, config: FrozenLakeEnvConfig = FrozenLakeEnvConfig()):
+         # Using mappings directly from config
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.GRID_LOOKUP = config.grid_lookup
+         self.ACTION_LOOKUP = config.action_lookup
+@@ -95,6 +96,7 @@ class FrozenLakeEnv(BaseDiscreteActionEnv, GymFrozenLakeEnv):
+ if __name__ == "__main__":
+     import matplotlib.pyplot as plt
+     config = FrozenLakeEnvConfig(size=4, is_slippery=True)
++    # import pdb;pdb.set_trace()
+     env = FrozenLakeEnv(config)
+     print(env.reset())
+     while True:
+diff --git a/ragen/env/sokoban/env.py b/ragen/env/sokoban/env.py
+index 17cd636..6922b54 100644
+--- a/ragen/env/sokoban/env.py
++++ b/ragen/env/sokoban/env.py
+@@ -33,6 +33,7 @@ class SokobanEnv(BaseDiscreteActionEnv, GymSokobanEnv):
+     def reset(self, seed=None, mode=None):
+         try:
+             with all_seed(seed):
++                # import pdb;pdb.set_trace()
+                 self.room_fixed, self.room_state, self.box_mapping, action_sequence = generate_room(
+                     dim=self.dim_room,
+                     num_steps=self.num_gen_steps,
+diff --git a/ragen/llm_agent/agent_proxy.py b/ragen/llm_agent/agent_proxy.py
+index c25b50d..f6c4088 100644
+--- a/ragen/llm_agent/agent_proxy.py
++++ b/ragen/llm_agent/agent_proxy.py
+@@ -287,6 +287,7 @@ def main(config):
+ 	proxy = LLMAgentProxy(config, actor_wg, tokenizer)
+ 	import time
+ 	start_time = time.time()
++	# import pdb;pdb.set_trace()
+ 	rollouts = proxy.rollout(
+ 		DataProto(
+ 			batch=None,
+diff --git a/ragen/llm_agent/base_llm.py b/ragen/llm_agent/base_llm.py
+index 358a2eb..b6f7797 100644
+--- a/ragen/llm_agent/base_llm.py
++++ b/ragen/llm_agent/base_llm.py
+@@ -33,6 +33,7 @@ class OpenAIProvider(LLMProvider):
+             raise ValueError("OpenAI API key not provided and not found in environment variables")
+         
+         self.client = AsyncOpenAI(api_key=self.api_key)
++        # import pdb;pdb.set_trace()
+     
+     async def generate(self, messages: List[Dict[str, str]], **kwargs) -> LLMResponse:
+         if "o1-mini" in self.model_name:
+@@ -46,10 +47,7 @@ class OpenAIProvider(LLMProvider):
+         )
+         if response.choices[0].finish_reason in ['length', 'content_filter']:
+             raise ValueError("Content filtered or length exceeded")
+-        return LLMResponse(
+-            content=response.choices[0].message.content,
+-            model_name=response.model
+-        )
++        return LLMResponse(content=response.choices[0].message.content,model_name=response.model)
+ 
+ class DeepSeekProvider(LLMProvider):
+     """DeepSeek API provider implementation"""
+@@ -199,7 +197,7 @@ class ConcurrentLLM:
+         
+         # Queue to store unfinished or failed tasks
+         current_batch = messages_list.copy()
+-        max_retries = kwargs.get("max_retries", 100)
++        max_retries = kwargs.get("max_retries", 10)
+         retry_count = 0
+         
+         while current_batch and retry_count < max_retries:
+diff --git a/ragen/llm_agent/ctx_manager.py b/ragen/llm_agent/ctx_manager.py
+index 905247a..af20a01 100644
+--- a/ragen/llm_agent/ctx_manager.py
++++ b/ragen/llm_agent/ctx_manager.py
+@@ -88,6 +88,7 @@ class ContextManager:
+         Initialize the ContextManager.
+         Processor is used to process the image data.
+         """
++        # import pdb;pdb.set_trace()
+         self.config = config
+         self.tokenizer = tokenizer
+         self.processor = processor
+@@ -311,7 +312,7 @@ class ContextManager:
+ 
+             llm_input_texts.append(text_with_prompt)
+             messages_list.append(messages)
+-
++        # import pdb;pdb.set_trace()
+         inputs = self.tokenizer(llm_input_texts, return_tensors="pt", padding=True, padding_side="left", truncation=False) # We have truncated previously, truncation in tokenizer may cause issues.
+         input_ids, attention_mask = inputs.input_ids, inputs.attention_mask
+         position_ids = (attention_mask.cumsum(dim=-1) - 1).clamp(min=0)
+@@ -353,6 +354,30 @@ class ContextManager:
+                 key: np.sum(value) / self.env_nums[key.split("/")[0]]
+                 for key, value in metrics.items()
+             }
++            # Derived metrics for wandb logging
++            try:
++                # charts/avg_episode_return: average across all 2048 env groups
++                ep_keys = [k for k in metrics.keys() if k.endswith('/episodic_return')]
++                if len(ep_keys) > 0:
++                    ep_vals = []
++                    for k in ep_keys:
++                        tag = k.split('/')[0]
++                        denom = self.env_nums.get(tag, max(1, len(metrics[k])))
++                        ep_vals.append(float(np.sum(metrics[k]) / denom))
++                    mean_metrics["charts/avg_episode_return"] = float(np.mean(ep_vals))
++            except Exception:
++                pass
++            try:
++                # rollout/max_tile: max over all envs in this batch
++                tile_keys = [k for k in metrics.keys() if k.endswith('/max_tile')]
++                if len(tile_keys) > 0:
++                    tile_vals = []
++                    for k in tile_keys:
++                        tile_vals.extend(metrics[k])
++                    if len(tile_vals) > 0:
++                        mean_metrics["rollout/max_tile"] = int(np.max(tile_vals))
++            except Exception:
++                pass
+             for key, values in metrics.items():
+                 if not isinstance(values, list):
+                     continue
+diff --git a/ragen/llm_agent/es_manager.py b/ragen/llm_agent/es_manager.py
+index b87a2b3..1a49d3e 100644
+--- a/ragen/llm_agent/es_manager.py
++++ b/ragen/llm_agent/es_manager.py
+@@ -128,18 +128,40 @@ class EnvStateManager:
+         env_outputs: List[Dict]
+             {env_id: int, history: List[Dict][{state: str, actions: List[str], reward: float, info: Dict, llm_response: str, llm_raw_response: str, (Optional)images: List[PIL.Image.Image]}]}
+         """
++        # def _execute_actions(env, actions):
++        #     acc_reward, turn_info, turn_done = 0, {}, False
++        #     executed_actions = []
++        #     for action in actions:
++        #         _, reward, done, info = env.step(action)
++        #         acc_reward += reward
++        #         turn_info.update(info) # NOTE: currently use last info for multi-action
++        #         executed_actions.append(action)
++        #         if done:
++        #             turn_done = True
++        #             break
++            
++        #     return acc_reward, turn_info, turn_done, executed_actions
+         def _execute_actions(env, actions):
+-            acc_reward, turn_info, turn_done = 0, {}, False
++            acc_reward, turn_info, turn_done = 0.0, {}, False
++            raw_acc_reward = 0.0
+             executed_actions = []
+             for action in actions:
+                 _, reward, done, info = env.step(action)
+-                acc_reward += reward
++                acc_reward += float(reward)
++                try:
++                    raw_acc_reward += float(info.get('raw_reward', 0.0))
++                except Exception:
++                    pass
+                 turn_info.update(info) # NOTE: currently use last info for multi-action
+                 executed_actions.append(action)
+                 if done:
+                     turn_done = True
+                     break
+-            
++            # Overwrite per-turn raw_reward to reflect the sum across all executed actions in this turn
++            try:
++                turn_info['raw_reward'] = float(raw_acc_reward)
++            except Exception:
++                pass
+             return acc_reward, turn_info, turn_done, executed_actions
+ 
+         def _log_env_state(status, history, cur_obs, max_actions_per_traj, executed_actions, all_actions, acc_reward, turn_done, turn_info, env_input):
+@@ -198,6 +220,18 @@ class EnvStateManager:
+                 'success': float(status.terminated and (not status.truncated)),
+                 'num_actions': status.num_actions,
+             }
++            # Add episodic-level metrics
++            # try:
++            #     # Sum of per-turn rewards equals the episodic return (env-shaped reward)
++            #     env_metric['episodic_return'] = float(sum(status.rewards))
++            # except Exception:
++            #     pass
++            try:
++                # Final max tile on the board at the end of the rollout
++                import numpy as _np
++                env_metric['max_tile'] = int(_np.max(entry['env'].grid))
++            except Exception:
++                pass
+             custom_metric = {}
+             for turn in cache['history']:
+                 for k, v in turn.get('info', {}).items():
+@@ -212,6 +246,12 @@ class EnvStateManager:
+                             "Skipping non-numeric metric '%s' with value %r for env %s.",
+                             k, v, entry['tag']
+                         )
++            # Add episodic_return as the SUM of raw_reward across steps (align with CleanRL)
++            try:
++                if 'raw_reward' in custom_metric:
++                    env_metric['episodic_return'] = float(np.sum(custom_metric['raw_reward']))
++            except Exception:
++                pass
+             for k, v in custom_metric.items():
+                 # TODO: Move TURN_LVL_METRICS into the environment
+                 if "webshop" not in cache['tag'].lower() or ("webshop" in cache['tag'].lower() and k in TURN_LVL_METRICS):
+@@ -219,7 +259,12 @@ class EnvStateManager:
+                 else:
+                     env_metric['traj_sum/' + k] = np.sum(v)
+ 
+-
++            try:
++                if 'score' in custom_metric and len(custom_metric['score']) > 0:
++                    env_metric['final_score'] = float(custom_metric['score'][-1])
++            except Exception:
++                pass
++                
+             cache['history'][-1]['metrics'] = custom_metric
+             env_metric = {f"{entry['tag']}/{k}": v for k, v in env_metric.items()}
+             cache['metrics'] = env_metric
+diff --git a/requirements.txt b/requirements.txt
+index 2fd756e..9b521bf 100644
+--- a/requirements.txt
++++ b/requirements.txt
+@@ -7,7 +7,6 @@ accelerate
+ codetiming
+ datasets
+ dill
+-flash-attn==2.7.4.post1
+ hydra-core
+ numpy
+ pandas
+@@ -15,19 +14,19 @@ pybind11
+ ray>=2.10
+ tensordict>=0.8.0,<0.9.0
+ transformers
+-vllm==0.8.2
++vllm==0.8.5
+ wandb
+ gymnasium
+ gymnasium[toy-text]
+ 
+ pyarrow>=15.0.0
+ pylatexenc
+-torchdata
++# torchdata
+ debugpy
+ 
+ together
+ anthropic
+-faiss-cpu==1.11.0
++faiss-gpu
+ 
+ # This is optional, but needs to be installed with main requirements if you want to use webshop
+ -r external/webshop-minimal/requirements.txt
+diff --git a/scripts/setup_ragen.sh b/scripts/setup_ragen.sh
+index f9a7cd9..85a93c4 100644
+--- a/scripts/setup_ragen.sh
++++ b/scripts/setup_ragen.sh
+@@ -94,10 +94,10 @@ main() {
+         pip install torch==2.5.0 --index-url https://download.pytorch.org/whl/cu124
+         
+         print_step "Installing flash-attention..."
+-        pip3 install flash-attn==2.7.4.post1 --no-build-isolation
++        # pip3 install flash-attn==2.7.4.post1 --no-build-isolation
+     else
+         print_step "Installing PyTorch without CUDA support..."
+-        pip install torch==2.5.0
++        pip install torch==2.4.0
+     fi
+     
+     # Install remaining requirements
+@@ -137,8 +137,8 @@ main() {
+     conda install conda-forge::gdown
+     mkdir -p external/webshop-minimal/webshop_minimal/data/full
+     cd external/webshop-minimal/webshop_minimal/data/full
+-    gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
+-    gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
++    # gdown https://drive.google.com/uc?id=1A2whVgOO0euk5O13n2iYDM0bQRkkRduB # items_shuffle
++    # gdown https://drive.google.com/uc?id=1s2j6NgHljiZzQNL3veZaAiyW_qDEgBNi # items_ins_v2
+     cd ../../../../..
+ 
+     echo -e "${GREEN}Installation completed successfully!${NC}"
+diff --git a/train_all.sh b/train_all.sh
+index 0157306..33035f5 100755
+--- a/train_all.sh
++++ b/train_all.sh
+@@ -6,246 +6,257 @@ USE_GRPO="algorithm.adv_estimator=grpo"
+ USE_PPO="algorithm.adv_estimator=gae" # by default.
+ USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+ 
++
++# python train.py --config-name _8_2048 system.CUDA_VISIBLE_DEVICES="'0,1,2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=game_2048 $USE_PPO $USE_BASE 
++
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-1 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-1.5B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++# python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'4,5'" trainer.project_name=ragen_latest_qwen_25_3b_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-3B-Instruct  trainer.experiment_name=rubikscube-2 $USE_PPO $USE_BASE 
++python train.py --config-name _10_rubikscube system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_05B_it trainer.n_gpus_per_node=2 model_path=Qwen/Qwen2.5-0.5B-Instruct  trainer.experiment_name=rubikscube-3 $USE_PPO $USE_BASE 
++
+ # Section 3.1&3.2 - General Observations
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozen_lake-ppo $USE_PPO $USE_BASE &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozen_lake-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'6,7'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=bandit-ppo-multitask $USE_PPO $USE_BASE &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="'7'" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-ppo-frommlp $USE_PPO $USE_BASE 
+ 
+-# Section 4.1 - Filtering and critic
+-# 0.25
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+-
+-wait
+-
+-# 0.5
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
+-
+-# 0.75
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+-
+-wait
+-
+-# Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="'0,1'" trainer.project_name=ragen_latest_qwen_25_15b_it  trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=sokoban-ppo-box1-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-grpo $USE_GRPO $USE_BASE &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'2,3'" trainer.project_name=ragen_latest_qwen_25_15b_it trainer.n_gpus_per_node=2 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-wait
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.project_name=ragen_latest_qwen_25_3b_it  trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_3B_it_multitask  trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="'4,5,6,7'" trainer.n_gpus_per_node=4 model_path=/mnt/general/wanghy/LLaMA-Factory/saves/qwen3/full/sft/qwen2.5_1.5B_it_multitask trainer.experiment_name=frozen_lake-ppo-slippery-multitask $USE_PPO $USE_BASE 
+ 
+-# Section 5.1 - Reasoning Helps Generalization
++# # Section 4.1 - Filtering and critic
++# # 0.25
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.25 actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.25 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
+-    agent_proxy.enable_think=False
++# # 0.5
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-ppo-rolloutfilter0.5 actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.5 $USE_GRPO &
++
++# # 0.75
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-ppo-rolloutfilter0.75 actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozen_lake-ppo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozen_lake-grpo actor_rollout_ref.rollout.rollout_filter_ratio=0.75 $USE_GRPO &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader"
++# wait
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
+-    custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
+-    custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
+-    custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
+-    custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
+-    agent_proxy.enable_think=False
++# # Section 4.2 - Ablation on Critic/ClipHigh/KL. Start from Basic and add more components. The best setting for StarPO in agent is rollout_filter+Critic+Cliphigh+NoKL
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=frozenlake-base-grpo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_GRPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=frozenlake-base-ppo algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=frozenlake-base-ppo-cliphigh algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.28 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=frozenlake-base-ppo-nokl algorithm.kl_ctrl.kl_coef=0.000 actor_rollout_ref.actor.kl_loss_coef=0.000 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1 $USE_PPO &
+ 
++# wait
+ 
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
++# # Section 5.1 - Reasoning Helps Generalization
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=bandit-generalization \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian"
+ 
+-# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=bandit-generalization-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Engineer" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Teacher" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Trader" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Librarian" \
++#     agent_proxy.enable_think=False
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=bandit-generalization-rev \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader"
+ 
+-# COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=bandit-generalization-rev-nothink \
++#     custom_envs.Bandit.env_config.lo_arm_name="Teacher" \
++#     custom_envs.Bandit.env_config.hi_arm_name="Engineer" \
++#     custom_envs.BanditTest.env_config.lo_arm_name="Librarian" \
++#     custom_envs.BanditTest.env_config.hi_arm_name="Trader" \
++#     agent_proxy.enable_think=False
+ 
+-wait
+ 
+-# Section 5.2 - what leads to better reasoning?
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization micro_batch_size_per_gpu=8 model_path=Qwen/Qwen2.5-1.5B-Instruct&
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-generalization-nothink $SOKOBAN_GENERALIZATION_CONFIG agent_proxy.enable_think=False &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
++# # SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=128 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SokobanDifferentGridVocab] es_manager.val.env_configs.n_groups=[128]"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization $SOKOBAN_GENERALIZATION_CONFIG &
+ 
+-wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
++# # COMPOSITIONALITY_CONFIG="es_manager.train.env_groups=16 es_manager.train.env_configs.tags=[Bandit,SimpleSokoban] es_manager.train.env_configs.n_groups=[8,8] es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[Bandit,SimpleSokoban,LargerSokoban,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128] actor_rollout_ref.rollout.rollout_filter_ratio=1" # NOTE that we don't filter out low-var rollout in this setting
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=compositional-generalization $COMPOSITIONALITY_CONFIG &
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=compositional-generalization-nothink $COMPOSITIONALITY_CONFIG agent_proxy.enable_think=False &
+ 
++# wait
+ 
+-wait
++# # Section 5.2 - what leads to better reasoning?
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B-Instruct &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct trainer.n_gpus_per_node=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-instruct $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B-Instruct trainer.n_gpus_per_node=4 &
+ 
+-# Section 6.1 varying action count
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
++# wait
+ 
+-# section 6.2 Varying prompt diversity
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-generalization-qwen2.5-0.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-0.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-1.5B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-3b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5,6,7\" trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-7B &
+ 
+-wait 
++# wait
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\"  trainer.n_gpus_per_node=4 trainer.n_gpus_per_node=4 trainer.experiment_name=sokoban-generalization-qwen2.5-7b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-7B &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\"  trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-generalization-qwen2.5-1.5b-r1 $SOKOBAN_GENERALIZATION_CONFIG model_path=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
++# wait
+ 
++# # Section 6.1 varying action count
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-action-count-1 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=5 custom_envs.LargerSokoban.max_actions_per_traj=5 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=5 custom_envs.FrozenLake.max_actions_per_traj=5 agent_proxy.max_actions_per_turn=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-action-count-2 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=10 custom_envs.LargerSokoban.max_actions_per_traj=10 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=10 custom_envs.FrozenLake.max_actions_per_traj=10 agent_proxy.max_actions_per_turn=2 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-action-count-3 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=15 custom_envs.LargerSokoban.max_actions_per_traj=15 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=15 custom_envs.FrozenLake.max_actions_per_traj=15 agent_proxy.max_actions_per_turn=3 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-action-count-4 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=20 custom_envs.LargerSokoban.max_actions_per_traj=20 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=20 custom_envs.FrozenLake.max_actions_per_traj=20 agent_proxy.max_actions_per_turn=4 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-action-count-5 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=25 custom_envs.LargerSokoban.max_actions_per_traj=25 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=25 custom_envs.FrozenLake.max_actions_per_traj=25 agent_proxy.max_actions_per_turn=5 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-action-count-6 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=30 custom_envs.LargerSokoban.max_actions_per_traj=30 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=30 custom_envs.FrozenLake.max_actions_per_traj=30 agent_proxy.max_actions_per_turn=6 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-action-count-7 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=35 custom_envs.LargerSokoban.max_actions_per_traj=35 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=35 custom_envs.FrozenLake.max_actions_per_traj=35 agent_proxy.max_actions_per_turn=7 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-action-count-8 $SOKOBAN_GENERALIZATION_CONFIG custom_envs.SimpleSokoban.max_actions_per_traj=40 custom_envs.LargerSokoban.max_actions_per_traj=40 custom_envs.SokobanDifferentGridVocab.max_actions_per_traj=40 custom_envs.FrozenLake.max_actions_per_traj=40 agent_proxy.max_actions_per_turn=8 &
+ 
++# # section 6.2 Varying prompt diversity
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-prompt-diversity-4 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=4 es_manager.train.group_size=32 es_manager.train.env_configs.n_groups=[4] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="1" trainer.experiment_name=sokoban-prompt-diversity-8 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="2" trainer.experiment_name=sokoban-prompt-diversity-16 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=8 es_manager.train.env_configs.n_groups=[16] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="3" trainer.experiment_name=sokoban-prompt-diversity-32 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=32 es_manager.train.group_size=4 es_manager.train.env_configs.n_groups=[32] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="4" trainer.experiment_name=sokoban-prompt-diversity-64 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=64 es_manager.train.group_size=2 es_manager.train.env_configs.n_groups=[64] &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="5" trainer.experiment_name=sokoban-prompt-diversity-128 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=128 es_manager.train.group_size=1 es_manager.train.env_configs.n_groups=[128] &
+ 
++# wait 
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="6" trainer.experiment_name=sokoban-online-2 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=16 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[16] trainer.total_training_steps=100 trainer.test_freq=5 &
+ 
+-# Extension: Training 7B reasoning model
+-SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="7" trainer.experiment_name=sokoban-online-5 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=40 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[40] trainer.total_training_steps=40 trainer.test_freq=2 &
+ 
+-python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES="0" trainer.experiment_name=sokoban-online-10 $SOKOBAN_GENERALIZATION_CONFIG es_manager.train.env_groups=80 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[80] trainer.total_training_steps=80 trainer.test_freq=1 &
+ 
+-# trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
+-python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
++# # Extension: Training 7B reasoning model
++# SOKOBAN_GENERALIZATION_CONFIG="es_manager.val.env_groups=512 es_manager.val.group_size=1 es_manager.val.env_configs.tags=[SimpleSokoban,LargerSokoban,SokobanDifferentGridVocab,FrozenLake] es_manager.val.env_configs.n_groups=[128,128,128,128]"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 trainer.experiment_name=sokoban-generalization-qwen2.5-3b-instruct-largescale $SOKOBAN_GENERALIZATION_CONFIG model_path=Qwen/Qwen2.5-3B-Instruct es_manager.train.env_groups=8 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[8] micro_batch_size_per_gpu=8 ppo_mini_batch_size=64 actor_rollout_ref.rollout.response_length=1024 actor_rollout_ref.rollout.max_model_len=6400 trainer.test_freq=5 actor_rollout_ref.rollout.max_num_batched_tokens=24000 micro_batch_size_per_gpu=2 actor_rollout_ref.rollout.rollout_filter_ratio=1 &
+ 
++# python -m ragen.llm_agent.agent_proxy  model_path=Qwen/Qwen2.5-3B-Instruct system.CUDA_VISIBLE_DEVICES=\"0,1,2,3,4,5,6,7\" trainer.n_gpus_per_node=8 actor_rollout_ref.rollout.tensor_model_parallel_size=4 actor_rollout_ref.rollout.response_length=2048 actor_rollout_ref.rollout.max_model_len=12800 
+ 
++# # trainer.save_freq=50 trainer.default_local_dir=/mnt/local/cache/exp_name
+ 
+-# USE_PPO="algorithm.adv_estimator=gae" # by default.
+-# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/sokoban-generalization &
+ 
++# python train.py --config-name _1_bandit system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=bandit-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/bandit-generalization &
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
++# python train.py --config-name _3_frozen_lake system.CUDA_VISIBLE_DEVICES=\"1\" trainer.n_gpus_per_node=1 trainer.experiment_name=frozenlake-final enable_response_mask=True trainer.total_training_steps=500 trainer.save_freq=50 trainer.default_local_dir=/mnt/local/ragen_checkpoints/frozenlake-generalization &
+ 
+-# enable_response_mask: False
+-# grpo_advantage_length_weight: True
+ 
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+ 
+-# extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
+-python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
++# # USE_PPO="algorithm.adv_estimator=gae" # by default.
++# # USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# # python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0\" trainer.n_gpus_per_node=1 trainer.experiment_name=sokoban-ppo $USE_PPO $USE_BASE ppo_mini_batch_size=64 enable_response_mask=True &
+ 
+-# extension: bi-level gae
+-python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
+-    system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
+-    model_path=Qwen/Qwen2.5-0.5B-Instruct \
+-    algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
+-    agent_proxy.use_turn_scores=True \
+-    actor_rollout_ref.rollout.tp_size_check=False
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std &
+ 
+-# extension: webshop
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # enable_response_mask: False
++# # grpo_advantage_length_weight: True
+ 
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 trainer.experiment_name=sokoban-s-grpo-1-5b algorithm.adv_estimator=grpo agent_proxy.reward_normalization.method=mean_std agent_proxy.max_actions_per_turn=5 custom_envs.SimpleSokoban.max_actions_per_traj=25 enable_response_mask=True grpo_advantage_length_weight=False model_path=Qwen/Qwen2.5-1.5B-Instruct &
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: 7B with lora. Currently NOT recommended to use lora: within current version of vllm, this could result in rollouts slower than non-lora by 100%
++# python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 model_path=Qwen/Qwen2.5-7B-Instruct trainer.experiment_name=sokoban_7b_instruct_lora_newversion lora.rank=16  
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # extension: bi-level gae
++# python train.py trainer.experiment_name=sokoban-bi-level-gae-final \
++#     system.CUDA_VISIBLE_DEVICES=\"0,1,2,3\" trainer.n_gpus_per_node=4 actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
++#     model_path=Qwen/Qwen2.5-0.5B-Instruct \
++#     algorithm.bi_level_gae=True algorithm.high_level_gamma=0.95 \
++#     agent_proxy.use_turn_scores=True \
++#     actor_rollout_ref.rollout.tp_size_check=False
+ 
+ 
+-# normal:sokoban
+-# extension: sokoban
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
+-    trainer.nnodes=1 &
++# # extension: webshop
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
+ 
+ 
+-USE_GRPO="algorithm.adv_estimator=grpo" # by default.
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
+-    trainer.nnodes=1 &
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
+ 
+-# StarPO ppo
+-USE_PPO="algorithm.adv_estimator=gae" # by default.
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+-# StarPO grpo
+-USE_GRPO="algorithm.adv_estimator=grpo"
+-USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
+-MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
+-    trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
+-    es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
+-    trainer.nnodes=1 &
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _6_webshop system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=webshop-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++
++# # normal:sokoban
++# # extension: sokoban
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo-s $USE_PPO \
++#     trainer.nnodes=1 &
++
++
++# USE_GRPO="algorithm.adv_estimator=grpo" # by default.
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"2,3\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo-s $USE_GRPO \
++#     trainer.nnodes=1 &
++
++# # StarPO ppo
++# USE_PPO="algorithm.adv_estimator=gae" # by default.
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"4,5\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-ppo $USE_PPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
++
++# # StarPO grpo
++# USE_GRPO="algorithm.adv_estimator=grpo"
++# USE_BASE="algorithm.kl_ctrl.kl_coef=0.001 actor_rollout_ref.actor.kl_loss_coef=0.001 actor_rollout_ref.actor.clip_ratio_high=0.2 actor_rollout_ref.rollout.rollout_filter_ratio=1"
++# MKL_SERVICE_FORCE_INTEL=1 python train.py --config-name _2_sokoban system.CUDA_VISIBLE_DEVICES=\"6,7\" trainer.n_gpus_per_node=2 \
++#     trainer.experiment_name=sokoban-3b-grpo $USE_GRPO $USE_BASE \
++#     es_manager.train.env_groups=2 es_manager.train.group_size=16 es_manager.train.env_configs.n_groups=[2] \
++#     trainer.nnodes=1 &
+ 
+ 
+ 
+-python train.py \
+-    trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+
++# python train.py \
++#     trainer.experiment_name=3b-full-ppo-test system.CUDA_VISIBLE_DEVICES=\"0,1\" trainer.n_gpus_per_node=2
+

wandb/run-20251218_185248-ts36d12x/files/requirements.txt

@@ -0,0 +1,315 @@
+colorama==0.4.6
+psutil==7.1.2
+ragen==0.1
+ragen==0.1
+setuptools==80.9.0
+wheel==0.45.1
+pip==25.3
+ragen==0.1
+pytz==2025.2
+nvidia-cusparselt-cu12==0.6.2
+mpmath==1.3.0
+typing_extensions==4.15.0
+sympy==1.13.1
+pylatexenc==2.10
+nvidia-cusolver-cu12==11.6.1.9
+numpy==1.26.4
+Markdown==3.10
+absl-py==2.3.1
+pyasn1_modules==0.4.2
+Werkzeug==3.1.3
+networkx==3.3
+MarkupSafe==2.1.5
+fsspec==2025.9.0
+filelock==3.19.1
+nvidia-cudnn-cu12==9.1.0.70
+Jinja2==3.1.6
+tensorboard==2.20.0
+triton==3.2.0
+pillow==11.3.0
+xxhash==3.6.0
+nvidia-nvtx-cu12==12.4.127
+nvidia-nvjitlink-cu12==12.4.127
+nvidia-nccl-cu12==2.21.5
+nvidia-curand-cu12==10.3.5.147
+wrapt==2.0.1
+nvidia-cufft-cu12==11.2.1.3
+pyjnius==1.7.0
+nvidia-cuda-runtime-cu12==12.4.127
+nvidia-cuda-nvrtc-cu12==12.4.127
+websockets==15.0.1
+nvidia-cuda-cupti-cu12==12.4.127
+zipp==3.23.0
+nvidia-cublas-cu12==12.4.5.8
+nvidia-cusparse-cu12==12.3.1.170
+flash_attn==2.8.3
+rsa==4.9.1
+einops==0.8.1
+py-cpuinfo==9.0.0
+pure_eval==0.2.3
+ptyprocess==0.7.0
+gym-notices==0.1.0
+flatbuffers==25.9.23
+fastrlock==0.8.3
+Farama-Notifications==0.0.4
+faiss-gpu==1.7.2
+antlr4-python3-runtime==4.9.3
+wcwidth==0.2.14
+wasabi==1.1.3
+uvloop==0.22.1
+urllib3==2.5.0
+tzdata==2025.2
+typing-inspection==0.4.2
+traitlets==5.14.3
+tqdm==4.67.1
+tomli==2.3.0
+threadpoolctl==3.6.0
+tenacity==9.1.2
+tabulate==0.9.0
+spacy-loggers==1.0.5
+spacy-legacy==3.0.12
+soupsieve==2.8
+sniffio==1.3.1
+smmap==5.0.2
+six==1.17.0
+shellingham==1.5.4
+sentencepiece==0.2.1
+scipy==1.15.3
+safetensors==0.6.2
+rpds-py==0.29.0
+rignore==0.7.6
+regex==2025.11.3
+RapidFuzz==3.14.3
+rank-bm25==0.2.2
+pyzmq==27.1.0
+PyYAML==6.0.3
+pytokens==0.3.0
+python-multipart==0.0.20
+python-json-logger==4.0.0
+python-dotenv==1.2.1
+PySocks==1.7.1
+pyparsing==3.2.5
+PyJWT==2.10.1
+Pygments==2.19.2
+pygame==2.6.1
+pydantic_core==2.41.5
+pycparser==2.23
+pycountry==24.6.1
+pybind11==3.0.1
+pyarrow==22.0.0
+psutil==7.1.3
+protobuf==4.25.8
+propcache==0.4.1
+prometheus_client==0.23.1
+platformdirs==4.5.0
+pexpect==4.9.0
+pathspec==0.12.1
+partial-json-parser==0.2.1.1.post6
+parso==0.8.5
+packaging==25.0
+orjson==3.11.4
+opentelemetry-semantic-conventions-ai==0.4.13
+opencv-python-headless==4.12.0.88
+ninja==1.13.0
+nest-asyncio==1.6.0
+mypy_extensions==1.1.0
+murmurhash==1.0.15
+multidict==6.7.0
+msgspec==0.19.0
+msgpack==1.1.2
+mdurl==0.1.2
+loguru==0.7.3
+llvmlite==0.44.0
+llguidance==0.7.30
+lark==1.2.2
+kiwisolver==1.4.9
+joblib==1.5.2
+jiter==0.12.0
+itsdangerous==2.2.0
+interegular==0.3.3
+ImageIO==2.37.2
+idna==3.11
+humanfriendly==10.0
+httpx-sse==0.4.3
+httptools==0.7.1
+html2text==2025.4.15
+hf-xet==1.2.0
+h11==0.16.0
+grpcio==1.76.0
+frozenlist==1.8.0
+fonttools==4.60.1
+executing==2.2.1
+exceptiongroup==1.3.0
+eval_type_backport==0.2.2
+docstring_parser==0.17.0
+dnspython==2.8.0
+distro==1.9.0
+diskcache==5.6.3
+dill==0.4.0
+decorator==5.2.1
+debugpy==1.8.17
+Cython==3.2.1
+cymem==2.0.13
+cycler==0.12.1
+cupy-cuda12x==13.6.0
+contourpy==1.3.2
+colorama==0.4.6
+codetiming==1.4.0
+cloudpickle==3.1.2
+cloudpathlib==0.23.0
+click==8.3.1
+charset-normalizer==3.4.4
+certifi==2025.11.12
+catalogue==2.0.10
+cachetools==6.2.2
+blis==1.3.2
+blinker==1.9.0
+blake3==1.0.8
+attrs==25.4.0
+async-timeout==5.0.1
+asttokens==3.0.1
+astor==0.8.1
+annotated-types==0.7.0
+annotated-doc==0.0.4
+airportsdata==20250909
+aiohappyeyeballs==2.6.1
+yarl==1.22.0
+uvicorn==0.38.0
+typer-slim==0.20.0
+thefuzz==0.22.1
+stack-data==0.6.3
+srsly==2.5.1
+smart_open==7.5.0
+sentry-sdk==2.44.0
+scikit-learn==1.7.2
+requests==2.32.5
+referencing==0.37.0
+python-dateutil==2.9.0.post0
+pydantic==2.12.4
+prompt_toolkit==3.0.52
+preshed==3.0.11
+opentelemetry-proto==1.26.0
+omegaconf==2.3.0
+numba==0.61.2
+nltk==3.9.2
+multiprocess==0.70.18
+matplotlib-inline==0.2.1
+markdown-it-py==4.0.0
+jedi==0.19.2
+importlib_metadata==8.0.0
+httpcore==1.0.9
+gymnasium==1.2.2
+gym==0.26.2
+googleapis-common-protos==1.72.0
+gitdb==4.0.12
+gguf==0.17.1
+Flask==3.1.2
+faiss-cpu==1.13.0
+email-validator==2.3.0
+depyf==0.18.0
+Deprecated==1.3.1
+coloredlogs==15.0.1
+cffi==2.0.0
+black==25.11.0
+beautifulsoup4==4.14.2
+anyio==4.11.0
+aiosignal==1.4.0
+watchfiles==1.1.1
+tiktoken==0.12.0
+starlette==0.49.3
+sse-starlette==3.0.3
+rich==14.2.0
+pydantic-settings==2.12.0
+pydantic-extra-types==2.10.6
+pandas==2.3.3
+opentelemetry-exporter-otlp-proto-common==1.26.0
+onnxruntime==1.23.2
+matplotlib==3.10.7
+lm-format-enforcer==0.10.12
+jsonschema-specifications==2025.9.1
+ipython==8.37.0
+hydra-core==1.3.2
+huggingface-hub==0.36.0
+httpx==0.28.1
+gym-sokoban==0.0.6
+GitPython==3.1.45
+cryptography==46.0.3
+confection==0.1.5
+cleantext==1.1.4
+aiohttp==3.13.2
+weasel==0.4.3
+wandb==0.23.0
+typer==0.19.2
+virtualenv==20.35.4
+pyasn1==0.6.1
+torch==2.6.0
+tokenizers==0.22.1
+thinc==8.3.9
+rich-toolkit==0.15.1
+prometheus-fastapi-instrumentator==7.1.0
+openai==2.8.0
+jsonschema==4.25.1
+gdown==5.2.0
+fastapi==0.121.2
+anthropic==0.73.0
+xformers==0.0.29.post2
+transformers==4.57.1
+proto-plus==1.26.1
+torchvision==0.21.0
+torchdata==0.11.0
+torchaudio==2.6.0
+together==1.5.30
+tensordict==0.8.3
+spacy==3.8.9
+ray==2.51.1
+outlines_core==0.1.26
+mistral_common==1.8.5
+mcp==1.21.1
+fastapi-cloud-cli==0.3.1
+fastapi-cli==0.0.16
+datasets==4.4.1
+accelerate==1.11.0
+xgrammar==0.1.18
+pyserini==1.2.0
+peft==0.18.0
+outlines==0.1.11
+opentelemetry-exporter-otlp-proto-http==1.26.0
+opentelemetry-exporter-otlp-proto-grpc==1.26.0
+kimina-client==0.2.1
+compressed-tensors==0.9.3
+opentelemetry-exporter-otlp==1.26.0
+vllm==0.8.5
+py-spy==0.4.1
+opencensus-context==0.1.3
+distlib==0.4.0
+colorful==0.5.8
+tensorboard-data-server==0.7.2
+google-auth==2.43.0
+google-api-core==2.28.1
+aiohttp-cors==0.8.1
+opentelemetry-exporter-prometheus==0.59b0
+opencensus==0.11.4
+verl==0.5.0.dev0
+opentelemetry-api==1.38.0
+opentelemetry-semantic-conventions==0.59b0
+opentelemetry-sdk==1.38.0
+typeguard==4.4.4
+shtab==1.8.0
+tyro==0.9.35
+verl==0.5.0.dev0
+autocommand==2.2.2
+backports.tarfile==1.2.0
+importlib_metadata==8.0.0
+inflect==7.3.1
+jaraco.collections==5.1.0
+jaraco.context==5.3.0
+jaraco.functools==4.0.1
+jaraco.text==3.12.1
+more-itertools==10.3.0
+packaging==24.2
+platformdirs==4.2.2
+tomli==2.0.1
+typeguard==4.3.0
+typing_extensions==4.12.2
+wheel==0.45.1
+zipp==3.19.2

wandb/run-20251218_185248-ts36d12x/files/wandb-metadata.json

@@ -0,0 +1,92 @@
+{
+  "os": "Linux-5.14.0-284.25.1.el9_2.x86_64-x86_64-with-glibc2.35",
+  "python": "CPython 3.10.19",
+  "startedAt": "2025-12-18T10:52:48.969828Z",
+  "args": [
+    "--track"
+  ],
+  "program": "/mnt/general/wanghy/RAGEN/cleanrl/cleanrl/ppo_rubikscube.py",
+  "codePath": "cleanrl/cleanrl/ppo_rubikscube.py",
+  "codePathLocal": "cleanrl/cleanrl/ppo_rubikscube.py",
+  "git": {
+    "remote": "https://github.com/mll-lab-nu/RAGEN.git",
+    "commit": "8d73639b99b38265453f898b8d6af7d4af50d56e"
+  },
+  "email": "haoyu-wa22@mails.tsinghua.edu.cn",
+  "root": "/mnt/general/wanghy/RAGEN",
+  "host": "pt-4d654cf4576f4d23ad3d3919f12932fe-worker-0",
+  "executable": "/root/local/miniconda3/envs/ragen/bin/python",
+  "cpu_count": 64,
+  "cpu_count_logical": 128,
+  "gpu": "NVIDIA H100 80GB HBM3",
+  "gpu_count": 8,
+  "disk": {
+    "/": {
+      "total": "5153960755200",
+      "used": "447561207808"
+    }
+  },
+  "memory": {
+    "total": "2163642122240"
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-35e2d43d-4067-82ce-90d4-def9e389bf28"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-af4135e3-88f2-e9ac-518d-502c75a85429"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-d7fdeeba-fe9b-ec03-d9f7-6724fe4266b5"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-ccc4f668-3882-5a8e-2c07-c5cd08f6f666"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-7b73c0cf-d3d5-e10c-7176-a43be1e41001"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-81b58d94-5d1f-8ec2-f9d2-fd56172ed177"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-03e8bc66-3b44-6794-49fd-5392fbdda6d1"
+    },
+    {
+      "name": "NVIDIA H100 80GB HBM3",
+      "memoryTotal": "85520809984",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-b7cf0ec6-7c29-1179-dceb-09565da51890"
+    }
+  ],
+  "cudaVersion": "12.4",
+  "writerId": "pa2plkmx4kakxqj0z9828uzvhc7cjmd1"
+}

wandb/run-20251218_185248-ts36d12x/files/wandb-summary.json

@@ -0,0 +1 @@
+{"rollout/ep_rew_mean":1,"_wandb":{"runtime":488},"losses/explained_variance":0.9999998941191635,"_timestamp":1.7660556590504558e+09,"train/old_approx_kl":1.4977995306253433e-05,"eval/std_return":0,"charts/SPS":2066,"train/clipfrac":0,"train/success_rate":1,"charts/avg_reward":0.5,"eval/episodes":4000,"global_step":999424,"eval/avg_return":1,"rollout/success_rate":1,"train/value_loss":1.2905440605059937e-12,"train/entropy":0.9372190237045288,"train/learning_rate":2.561475409836034e-07,"_step":999424,"losses/value_loss":1.2905440605059937e-12,"_runtime":488.947236407,"losses/policy_loss":-2.1204352378845215e-05,"eval/success_rate":1,"train/approx_kl":5.471520125865936e-08,"perf/SPS":2066,"charts/train_success_rate":1,"rollout/ep_len_mean":1,"charts/avg_value":0.9950001835823059,"train/policy_loss":-2.1204352378845215e-05,"charts/progress":100}

wandb/run-20251218_185248-ts36d12x/run-ts36d12x.wandb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4130893ad88b9d27d5be99b364340162db95ae9e6a5388fa9dfc4d17a2a17109
+size 57189556

wandb/run-20251218_190243-6eiq1qnz/files/config.yaml

@@ -0,0 +1,961 @@
+_wandb:
+    value:
+        cli_version: 0.23.0
+        e:
+            ksnadu1wq1hn3j4tn7kvyr37dur8onv0:
+                args:
+                    - --node-ip-address=10.119.16.10
+                    - --node-manager-port=46457
+                    - --object-store-name=/tmp/ray/session_2025-12-18_19-01-25_051217_50/sockets/plasma_store
+                    - --raylet-name=/tmp/ray/session_2025-12-18_19-01-25_051217_50/sockets/raylet
+                    - --redis-address=None
+                    - --metrics-agent-port=59353
+                    - --logging-rotate-bytes=536870912
+                    - --logging-rotate-backup-count=5
+                    - --runtime-env-agent-port=56001
+                    - --gcs-address=10.119.16.10:49077
+                    - --session-name=session_2025-12-18_19-01-25_051217_50
+                    - --temp-dir=/tmp/ray
+                    - --webui=127.0.0.1:8265
+                    - --cluster-id=f44abd8a0353dce8bb3cfc59c203d66bdcc2d6cc10d0841735322fe3
+                    - --startup-token=112
+                    - --worker-launch-time-ms=1766055688989
+                    - --node-id=246330e3702e1feff374f74c418a746f3fd74a876302f01bd5b8d5b4
+                    - --runtime-env-hash=-839022692
+                cpu_count: 64
+                cpu_count_logical: 128
+                cudaVersion: "12.4"
+                disk:
+                    /:
+                        total: "5153960755200"
+                        used: "3194880"
+                email: haoyu-wa22@mails.tsinghua.edu.cn
+                executable: /root/local/miniconda3/envs/ragen/bin/python
+                git:
+                    commit: 8d73639b99b38265453f898b8d6af7d4af50d56e
+                    remote: https://github.com/mll-lab-nu/RAGEN.git
+                gpu: NVIDIA H100 80GB HBM3
+                gpu_count: 8
+                gpu_nvidia:
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-34b1ceb3-41bf-8b15-936b-0cf63020796b
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-bb0e21ee-96da-91a4-7dde-5f2dada5f894
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-8788342f-112d-c7cf-f513-0bf789f08656
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-bb645fc4-7841-b48c-ba0c-75967eeaa342
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-85d58b2b-da3c-6a28-db10-99f31a8901dd
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-3744bc51-7758-b4d2-0df8-fd7c43393254
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-784deca5-2373-020f-5ee6-f841f00d6f77
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "85520809984"
+                      name: NVIDIA H100 80GB HBM3
+                      uuid: GPU-ea6ca467-24fc-212b-7fc3-2bd657db30c1
+                host: pt-9f0a38817de94c12810804da51920883-worker-0
+                memory:
+                    total: "2163639472128"
+                os: Linux-5.14.0-284.25.1.el9_2.x86_64-x86_64-with-glibc2.35
+                program: /root/local/miniconda3/envs/ragen/lib/python3.10/site-packages/ray/_private/workers/default_worker.py
+                python: CPython 3.10.19
+                root: /mnt/general/wanghy/RAGEN
+                startedAt: "2025-12-18T11:02:43.262285Z"
+                writerId: ksnadu1wq1hn3j4tn7kvyr37dur8onv0
+        m: []
+        python_version: 3.10.19
+        t:
+            "1":
+                - 1
+                - 5
+                - 11
+                - 30
+                - 41
+                - 49
+                - 50
+                - 51
+                - 53
+                - 71
+                - 95
+                - 97
+                - 98
+                - 105
+            "2":
+                - 1
+                - 5
+                - 11
+                - 30
+                - 41
+                - 49
+                - 50
+                - 51
+                - 53
+                - 71
+                - 95
+                - 97
+                - 98
+                - 105
+            "3":
+                - 2
+                - 13
+                - 16
+                - 61
+            "4": 3.10.19
+            "5": 0.23.0
+            "6": 4.57.3
+            "12": 0.23.0
+            "13": linux-x86_64
+actor_rollout_ref:
+    value:
+        actor:
+            _target_: verl.workers.config.FSDPActorConfig
+            checkpoint:
+                _target_: verl.trainer.config.CheckpointConfig
+                async_save: false
+                load_contents:
+                    - model
+                    - optimizer
+                    - extra
+                save_contents:
+                    - model
+                    - optimizer
+                    - extra
+            clip_ratio: 0.2
+            clip_ratio_c: 3
+            clip_ratio_high: 0.28
+            clip_ratio_low: 0.2
+            entropy_checkpointing: false
+            entropy_coeff: 0.001
+            entropy_from_logits_with_chunking: false
+            freeze_vision_tower: false
+            fsdp_config:
+                _target_: verl.workers.config.FSDPEngineConfig
+                entropy_checkpointing: false
+                entropy_from_logits_with_chunking: false
+                forward_only: false
+                forward_prefetch: false
+                fsdp_size: -1
+                model_dtype: fp32
+                offload_policy: false
+                optimizer_offload: false
+                param_offload: false
+                reshard_after_forward: true
+                strategy: fsdp
+                ulysses_sequence_parallel_size: 1
+                use_orig_params: false
+                use_torch_compile: true
+                wrap_policy:
+                    min_num_params: 0
+            grad_clip: 1
+            grpo_advantage_length_weight: false
+            kl_loss_coef: 0
+            kl_loss_type: kl
+            loss_agg_mode: token-mean
+            optim:
+                _target_: verl.workers.config.FSDPOptimizerConfig
+                betas:
+                    - 0.9
+                    - 0.999
+                clip_grad: 1
+                lr: 1e-06
+                lr_warmup_steps: -1
+                lr_warmup_steps_ratio: 0
+                min_lr_ratio: 0
+                num_cycles: 0.5
+                total_training_steps: 200
+                warmup_style: constant
+                weight_decay: 0.01
+            policy_loss:
+                _target_: verl.workers.config.PolicyLossConfig
+                clip_cov_lb: 1
+                clip_cov_ratio: 0.0002
+                clip_cov_ub: 5
+                kl_cov_ratio: 0.0002
+                loss_mode: vanilla
+                ppo_kl_coef: 0.1
+            ppo_epochs: 1
+            ppo_max_token_len_per_gpu: 16384
+            ppo_micro_batch_size: null
+            ppo_micro_batch_size_per_gpu: 1
+            ppo_mini_batch_size: 32
+            profiler:
+                _target_: verl.utils.profiler.ProfilerConfig
+                all_ranks: false
+                enable: false
+                ranks: []
+                save_path: outputs/profile
+                tool: null
+                tool_config:
+                    npu:
+                        _target_: verl.utils.profiler.config.NPUToolConfig
+                        analysis: true
+                        contents: []
+                        discrete: false
+                        level: level1
+                    nsys:
+                        _target_: verl.utils.profiler.config.NsightToolConfig
+                        discrete: false
+                    torch:
+                        _target_: verl.utils.profiler.config.TorchProfilerToolConfig
+                        step_end: null
+                        step_start: 0
+                    torch_memory:
+                        _target_: verl.utils.profiler.config.TorchMemoryToolConfig
+                        stack_depth: 32
+                        trace_alloc_max_entries: 100000
+            shuffle: false
+            strategy: fsdp
+            tis_imp_ratio_cap: -1
+            ulysses_sequence_parallel_size: 1
+            use_dynamic_bsz: false
+            use_fused_kernels: false
+            use_kl_loss: false
+            use_ref: true
+            use_remove_padding: false
+            use_torch_compile: true
+        hybrid_engine: true
+        model:
+            _target_: verl.workers.config.HFModelConfig
+            custom_chat_template: null
+            enable_activation_offload: false
+            enable_gradient_checkpointing: true
+            exclude_modules: null
+            external_lib: null
+            fused_kernel_options:
+                impl_backend: torch
+            hf_config_path: null
+            lora_alpha: 16
+            lora_rank: 0
+            path: Qwen/Qwen2.5-0.5B-Instruct
+            target_modules: all-linear
+            tokenizer_path: null
+            trust_remote_code: false
+            use_fused_kernels: false
+            use_liger: false
+            use_remove_padding: false
+            use_shm: false
+        nccl_timeout: 600
+        ref:
+            entropy_checkpointing: false
+            entropy_from_logits_with_chunking: false
+            fsdp_config:
+                _target_: verl.workers.config.FSDPEngineConfig
+                entropy_checkpointing: false
+                entropy_from_logits_with_chunking: false
+                forward_only: false
+                forward_prefetch: false
+                fsdp_size: -1
+                model_dtype: fp32
+                offload_policy: false
+                optimizer_offload: false
+                param_offload: false
+                reshard_after_forward: true
+                strategy: fsdp
+                ulysses_sequence_parallel_size: 1
+                use_orig_params: false
+                use_torch_compile: true
+                wrap_policy:
+                    min_num_params: 0
+            log_prob_max_token_len_per_gpu: 16384
+            log_prob_micro_batch_size: null
+            log_prob_micro_batch_size_per_gpu: 1
+            log_prob_use_dynamic_bsz: false
+            model: null
+            profiler:
+                _target_: verl.utils.profiler.ProfilerConfig
+                all_ranks: false
+                enable: false
+                ranks: []
+                save_path: outputs/profile
+                tool: null
+                tool_config:
+                    npu:
+                        _target_: verl.utils.profiler.config.NPUToolConfig
+                        analysis: true
+                        contents: []
+                        discrete: false
+                        level: level1
+                    nsys:
+                        _target_: verl.utils.profiler.config.NsightToolConfig
+                        discrete: false
+                    torch:
+                        _target_: verl.utils.profiler.config.TorchProfilerToolConfig
+                        step_end: null
+                        step_start: 0
+                    torch_memory:
+                        _target_: verl.utils.profiler.config.TorchMemoryToolConfig
+                        stack_depth: 32
+                        trace_alloc_max_entries: 100000
+            strategy: fsdp
+            ulysses_sequence_parallel_size: 1
+            use_torch_compile: true
+        rollout:
+            _target_: verl.workers.config.RolloutConfig
+            agent:
+                _target_: verl.workers.config.AgentLoopConfig
+                agent_loop_config_path: null
+                custom_async_server:
+                    _target_: verl.workers.config.CustomAsyncServerConfig
+                    name: null
+                    path: null
+                num_workers: 8
+            calculate_log_probs: false
+            cudagraph_capture_sizes: null
+            data_parallel_size: 1
+            disable_log_stats: true
+            do_sample: true
+            dtype: bfloat16
+            enable_chunked_prefill: true
+            enable_prefix_caching: true
+            enforce_eager: true
+            expert_parallel_size: 1
+            free_cache_engine: true
+            gpu_memory_utilization: 0.7
+            ignore_eos: false
+            layered_summon: false
+            load_format: dummy
+            log_prob_max_token_len_per_gpu: 16384
+            log_prob_micro_batch_size: null
+            log_prob_micro_batch_size_per_gpu: 1
+            log_prob_use_dynamic_bsz: false
+            max_model_len: 7200
+            max_num_batched_tokens: 8192
+            max_num_seqs: 1024
+            mode: sync
+            multi_stage_wake_up: false
+            multi_turn:
+                _target_: verl.workers.config.MultiTurnConfig
+                enable: false
+                format: hermes
+                interaction_config_path: null
+                max_assistant_turns: null
+                max_parallel_calls: 1
+                max_tool_response_length: 256
+                max_user_turns: null
+                num_repeat_rollouts: null
+                tokenization_sanity_check_mode: strict
+                tool_config_path: null
+                tool_response_truncate_side: middle
+                use_inference_chat_template: false
+            "n": 1
+            name: vllm
+            over_sample_rate: 0
+            profiler:
+                _target_: verl.utils.profiler.ProfilerConfig
+                all_ranks: false
+                enable: false
+                ranks: []
+                save_path: outputs/profile
+                tool: null
+                tool_config:
+                    npu:
+                        _target_: verl.utils.profiler.config.NPUToolConfig
+                        analysis: true
+                        contents: []
+                        discrete: false
+                        level: level1
+                    nsys:
+                        _target_: verl.utils.profiler.config.NsightToolConfig
+                        discrete: false
+                    torch:
+                        _target_: verl.utils.profiler.config.TorchProfilerToolConfig
+                        step_end: null
+                        step_start: 0
+                    torch_memory:
+                        _target_: verl.utils.profiler.config.TorchMemoryToolConfig
+                        stack_depth: 32
+                        trace_alloc_max_entries: 100000
+            prompt_length: 1
+            response_length: 500
+            rollout_filter_metric: reward_variance
+            rollout_filter_ratio: 0.25
+            rollout_filter_type: largest
+            skip_dump_dir: /tmp/rollout_dump
+            skip_rollout: false
+            skip_tokenizer_init: true
+            temperature: 1
+            tensor_model_parallel_size: 1
+            top_k: -1
+            top_p: 1
+            trace:
+                _target_: verl.workers.config.TraceConfig
+                backend: null
+                token2text: false
+            update_weights_bucket_megabytes: 512
+            val_kwargs:
+                _target_: verl.workers.config.SamplingConfig
+                do_sample: true
+                "n": 1
+                temperature: 0.5
+                top_k: -1
+                top_p: 1
+agent_proxy:
+    value:
+        action_sep: '||'
+        enable_think: true
+        max_actions_per_turn: 4
+        max_context_window: 5
+        max_turn: 25
+        reward_normalization:
+            grouping: state
+            method: identity
+        use_turn_scores: false
+algorithm:
+    value:
+        _target_: verl.trainer.config.AlgoConfig
+        adv_estimator: gae
+        bi_level_gae: false
+        gamma: 1
+        high_level_gamma: 0.95
+        kl_ctrl:
+            _target_: verl.trainer.config.KLControlConfig
+            horizon: 10000
+            kl_coef: 0
+            target_kl: 0.1
+            type: fixed
+        kl_penalty: kl
+        lam: 1
+        norm_adv_by_std_in_grpo: true
+        pf_ppo:
+            reweight_method: pow
+            weight_pow: 2
+        use_kl_in_reward: false
+        use_pf_ppo: false
+critic:
+    value:
+        _target_: verl.workers.config.FSDPCriticConfig
+        checkpoint:
+            _target_: verl.trainer.config.CheckpointConfig
+            async_save: false
+            load_contents:
+                - model
+                - optimizer
+                - extra
+            save_contents:
+                - model
+                - optimizer
+                - extra
+        cliprange_value: 0.5
+        enable: null
+        forward_max_token_len_per_gpu: 32768
+        forward_micro_batch_size: null
+        forward_micro_batch_size_per_gpu: 1
+        grad_clip: 1
+        loss_agg_mode: token-mean
+        model:
+            _target_: verl.workers.config.FSDPCriticModelCfg
+            enable_activation_offload: false
+            enable_gradient_checkpointing: true
+            external_lib: null
+            fsdp_config:
+                _target_: verl.workers.config.FSDPEngineConfig
+                entropy_checkpointing: false
+                entropy_from_logits_with_chunking: false
+                forward_only: false
+                forward_prefetch: false
+                fsdp_size: -1
+                model_dtype: fp32
+                offload_policy: false
+                optimizer_offload: false
+                param_offload: false
+                reshard_after_forward: true
+                strategy: fsdp
+                ulysses_sequence_parallel_size: 1
+                use_orig_params: false
+                use_torch_compile: true
+                wrap_policy:
+                    min_num_params: 0
+            lora_alpha: 16
+            lora_rank: 0
+            path: Qwen/Qwen2.5-0.5B-Instruct
+            target_modules: all-linear
+            tokenizer_path: Qwen/Qwen2.5-0.5B-Instruct
+            trust_remote_code: false
+            use_remove_padding: false
+            use_shm: false
+        optim:
+            _target_: verl.workers.config.FSDPOptimizerConfig
+            betas:
+                - 0.9
+                - 0.999
+            clip_grad: 1
+            lr: 1e-05
+            lr_warmup_steps: -1
+            lr_warmup_steps_ratio: 0
+            min_lr_ratio: 0
+            num_cycles: 0.5
+            total_training_steps: 200
+            warmup_style: constant
+            weight_decay: 0.01
+        ppo_epochs: 1
+        ppo_max_token_len_per_gpu: 32768
+        ppo_micro_batch_size: null
+        ppo_micro_batch_size_per_gpu: 1
+        ppo_mini_batch_size: 32
+        profiler:
+            _target_: verl.utils.profiler.ProfilerConfig
+            all_ranks: false
+            enable: false
+            ranks: []
+            save_path: outputs/profile
+            tool: null
+            tool_config:
+                npu:
+                    _target_: verl.utils.profiler.config.NPUToolConfig
+                    analysis: true
+                    contents: []
+                    discrete: false
+                    level: level1
+                nsys:
+                    _target_: verl.utils.profiler.config.NsightToolConfig
+                    discrete: false
+                torch:
+                    _target_: verl.utils.profiler.config.TorchProfilerToolConfig
+                    step_end: null
+                    step_start: 0
+                torch_memory:
+                    _target_: verl.utils.profiler.config.TorchMemoryToolConfig
+                    stack_depth: 32
+                    trace_alloc_max_entries: 100000
+        rollout_n: 1
+        shuffle: false
+        strategy: fsdp
+        ulysses_sequence_parallel_size: 1
+        use_dynamic_bsz: false
+ctx_manager:
+    value:
+        generation:
+            gen_config:
+                kwargs: null
+                response_length: 500
+                temperature: 1
+                top_k: -1
+                top_p: 1
+custom_envs:
+    value:
+        Bandit:
+            env_config:
+                split: train
+            env_instruction: ""
+            env_type: bandit
+            max_actions_per_traj: 1
+            max_tokens: 100
+        BanditTest:
+            env_config:
+                split: test
+            env_instruction: ""
+            env_type: bandit
+            max_actions_per_traj: 1
+            max_tokens: 100
+        CoordFrozenLake:
+            env_config:
+                observation_format: grid_coord
+            env_instruction: |
+                You are solving the FrozenLake puzzle. The observation includes both a symbol grid and zero-indexed coordinates for the start, goal, player, and any holes.
+                Coordinates range from the top-left corner (0, 0) to the bottom-right corner (5, 5).
+                Beware that the ice is slippery, so the agent might slide and end up in an unintended tile.
+                Respond with a sequence of actions such as <answer>Left || Up || Up</answer>.
+            env_type: frozen_lake
+            max_actions_per_traj: 25
+            max_tokens: 120
+        CoordSokoban:
+            env_config:
+                dim_x: 6
+                dim_y: 6
+                max_steps: 100
+                num_boxes: 2
+                observation_format: grid_coord
+            env_instruction: "You are solving the Sokoban puzzle. You are the player and you need to push all boxes to targets.\nYou are provided with a symbol grid and the zero-indexed coordinates of the player, each box, and each target. \nCoordinates range from the top-left corner (0, 0) to the bottom-right corner (5, 5). \nWhen you are exactly next to a box, you can push it by moving in the same direction. \nYou cannot push a box through a wall, and you cannot pull a box.\nThe answer should be a sequence of actions, like <answer>Right || Right || Up</answer>.\n"
+            env_type: sokoban
+            max_actions_per_traj: 10
+            max_tokens: 120
+        Countdown:
+            env_config: null
+            env_instruction: 'You are solving the Countdown puzzle. You should use the num list to create an equation that equals the target. Example answer format: <think> To find an equation using [3, 5, 2] to get 4. Let''s check 2 + 5 = 7, 7 - 3 = 4. So the answer is 2 + 5 - 3 = 4. </think><answer>2 + 5 - 3</answer>'
+            env_type: countdown
+            max_actions_per_traj: 1
+            max_tokens: 100
+        FrozenLake:
+            env_config: null
+            env_instruction: 'You are solving the FrozenLake puzzle. Forbid the whole and go to the target. You may move to the unintended direction due to the slippery ice. Example answer format: <think>To forbid the hole and go to the target, I should go left then go up.</think><answer>Left || Up</answer>'
+            env_type: frozen_lake
+            max_actions_per_traj: 25
+            max_tokens: 100
+        LargerSokoban:
+            env_config:
+                dim_x: 8
+                dim_y: 8
+                max_steps: 100
+                num_boxes: 2
+                search_depth: 10
+            env_instruction: "You are solving the Sokoban puzzle. \nYou are the player and you need to push all boxes to targets. \nWhen you are right next to a box, you can push it by moving in the same direction. \nYou cannot push a box through a wall, and you cannot pull a box. \nThe answer should be a sequence of actions, like <answer>Right || Right || Up</answer>\n"
+            env_type: sokoban
+            max_actions_per_traj: 10
+            max_tokens: 100
+        Lean:
+            env_config: null
+            env_instruction: You are a Lean theorem prover. Given a Lean theorem statement, propose a sequence of tactics that completes the proof. Think step by step about which tactics to apply next. Provide tactics separated by '||', for example <answer>intro || simp || rfl</answer>.
+            env_type: lean
+            max_actions_per_traj: 30
+            max_tokens: 512
+        MediumSudoku:
+            env_config:
+                difficulty: medium
+                grid_size: 9
+                max_steps: 81
+                render_format: with_feedback
+                show_conflicts: true
+                show_valid_numbers: true
+            env_instruction: |
+                You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 3x3 box contains the numbers 1-9 without repetition.
+                Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
+                Place numbers one at a time using the format: <answer>place 5 at row 2 col 3</answer> or <answer>2,3,5</answer>
+                The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
+            env_type: sudoku
+            max_actions_per_traj: 30
+            max_tokens: 150
+            max_workers: 32
+            parallel_friendly: false
+        MetamathQA:
+            env_config: null
+            env_instruction: 'You are solving Math problems. '
+            env_type: metamathqa
+            max_actions_per_traj: 1
+            max_tokens: 100
+        SimpleSokoban:
+            env_config:
+                dim_x: 6
+                dim_y: 6
+                max_steps: 100
+                num_boxes: 2
+            env_instruction: "You are solving the Sokoban puzzle. \nYou are the player and you need to push all boxes to targets. \nWhen you are right next to a box, you can push it by moving in the same direction. \nYou cannot push a box through a wall, and you cannot pull a box. \nThe answer should be a sequence of actions, like <answer>Right || Right || Up</answer>\n"
+            env_type: sokoban
+            max_actions_per_traj: 10
+            max_tokens: 100
+        SimpleSudoku:
+            env_config:
+                difficulty: easy
+                grid_size: 4
+                max_steps: 20
+                render_format: with_feedback
+                show_conflicts: false
+                show_valid_numbers: false
+            env_instruction: |
+                You are solving a Sudoku puzzle. Fill in the grid so that every row, column, and 2x2 box contains the numbers 1-4 without repetition.
+                Initial cells are shown in [brackets] and cannot be modified. Empty cells are shown as dots (.).
+                Place numbers one at a time using the format, for example: <answer>place 1 at row 2 col 3</answer> or <answer>1,2,3</answer>
+                The environment will provide feedback on valid/invalid moves and show conflicts if any occur.
+            env_type: sudoku
+            max_actions_per_traj: 20
+            max_tokens: 150
+            max_workers: 32
+            parallel_friendly: false
+        SokobanDifferentGridVocab:
+            env_config:
+                dim_x: 6
+                dim_y: 6
+                grid_lookup:
+                    "0": W
+                    "1": .
+                    "2": G
+                    "3": C
+                    "4": B
+                    "5": A
+                    "6": '@'
+                grid_vocab:
+                    .: empty
+                    '@': player on target
+                    A: player
+                    B: box
+                    C: box on target
+                    G: target
+                    W: wall
+                max_steps: 100
+                num_boxes: 1
+                search_depth: 30
+            env_instruction: "You are solving the Sokoban puzzle. \nYou are the player and you need to push all boxes to targets. \nWhen you are right next to a box, you can push it by moving in the same direction. \nYou cannot push a box through a wall, and you cannot pull a box. \nThe answer should be a sequence of actions, like <answer>Right || Right || Up</answer>\n"
+            env_type: sokoban
+            max_actions_per_traj: 10
+            max_tokens: 100
+        VisualSimpleSokoban:
+            env_config:
+                dim_x: 6
+                dim_y: 6
+                max_steps: 100
+                num_boxes: 1
+                render_mode: rgb_array
+            env_instruction: You are solving the Sokoban puzzle. You are the player and you need to push all boxes to targets. When you are right next to a box, you can push it by moving in the same direction. You cannot push a box through a wall, and you cannot pull a box. The answer should be a sequence of actions, like <answer>Right || Right || Up</answer>
+            env_type: sokoban
+            max_actions_per_traj: 10
+            max_tokens: 100
+        WebShop:
+            env_config:
+                dataset: small
+            env_instruction: 'You are browsing an online shop. Based on the instruction, buy a product that close to the production description. You need to search, read the search results, pick a product, choose the size and color and buy. You should only choose action from the available actions list provided later.  Example process: I need a gingko light and 20x20 pillow cover that is hand painted. First search[gingko light 20x20 pillow cover hand painted], answer format: <answer>search[blanket with fleece throw]</answer>. Valid answer is search[<keywords>] or click[<clickable>].'
+            env_type: webshop
+            max_actions_per_traj: 9
+            max_tokens: 200
+        blackjack:
+            env_config: null
+            env_instruction: |
+                You are playing Blackjack against a dealer. The dealer must hit on 16 or less and stand on 17 or more.
+                Choose either Stick or Hit. Respond with a single action.
+                Example: <answer>Hit</answer>
+            env_type: blackjack
+            max_actions_per_traj: 10
+            max_tokens: 64
+        game_2048:
+            env_config: null
+            env_instruction: |
+                You are playing the 2048 game on a 4x4 grid. Merge equal tiles by sliding Up, Right, Down, or Left.
+                If a move is invalid (no tiles move), a small penalty is applied. Respond with a single action.
+                Example: <answer>Up</answer>
+            env_type: game_2048
+            max_actions_per_traj: 700
+            max_tokens: 8192
+        rubikscube:
+            env_config:
+                max_steps: 20
+                render_mode: text
+                scramble_depth: 1
+            env_instruction: |
+                You are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.
+                Available actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.
+                - Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).
+                - Modifiers: A letter alone means 90° clockwise (e.g., 'R'). A letter with prime (') means 90° counter-clockwise (e.g., "R'").
+                Respond with a sequence of actions separated by "||".
+                Example: <answer>U</answer>
+            env_type: rubikscube
+            max_actions_per_traj: 20
+            max_tokens: 96
+custom_reward_function:
+    value:
+        name: compute_score
+        path: null
+data:
+    value:
+        custom_cls:
+            name: null
+            path: null
+        datagen:
+            name: null
+            path: null
+        dataloader_num_workers: 8
+        filter_overlong_prompts: false
+        filter_overlong_prompts_workers: 1
+        image_key: images
+        max_prompt_length: null
+        max_response_length: null
+        prompt_key: prompt
+        return_full_prompt: false
+        return_multi_modal_inputs: true
+        return_raw_chat: false
+        return_raw_input_ids: false
+        reward_fn_key: data_source
+        sampler:
+            class_name: null
+            class_path: null
+        shuffle: true
+        tokenizer: null
+        train_batch_size: 128
+        train_files: ~/data/rlhf/gsm8k/train.parquet
+        truncation: error
+        trust_remote_code: false
+        use_shm: false
+        val_batch_size: null
+        val_files: ~/data/rlhf/gsm8k/test.parquet
+        validation_shuffle: false
+        video_key: videos
+enable_response_mask:
+    value: true
+es_manager:
+    value:
+        format_penalty: -0.1
+        train:
+            env_configs:
+                n_groups:
+                    - 8
+                tags:
+                    - rubikscube
+            env_groups: 8
+            group_size: 16
+        val:
+            env_configs:
+                n_groups:
+                    - 256
+                tags:
+                    - rubikscube
+            env_groups: 256
+            group_size: 1
+global_profiler:
+    value:
+        _target_: verl.utils.profiler.ProfilerConfig
+        global_tool_config:
+            nsys:
+                _target_: verl.utils.profiler.config.NsightToolConfig
+                controller_nsight_options:
+                    cuda-graph-trace: graph
+                    cuda-memory-usage: "true"
+                    trace: cuda,nvtx,cublas,ucx
+                discrete: false
+                worker_nsight_options:
+                    capture-range: cudaProfilerApi
+                    capture-range-end: null
+                    cuda-graph-trace: graph
+                    cuda-memory-usage: "true"
+                    kill: none
+                    trace: cuda,nvtx,cublas,ucx
+            torch_memory:
+                context: all
+                stack_depth: 32
+                stacks: all
+                trace_alloc_max_entries: 100000
+        profile_continuous_steps: false
+        save_path: outputs/profile
+        steps: null
+        tool: null
+grpo_advantage_length_weight:
+    value: false
+lora:
+    value:
+        alpha: 16
+        rank: 0
+        target_modules: all-linear
+micro_batch_size_per_gpu:
+    value: 1
+model_path:
+    value: Qwen/Qwen2.5-0.5B-Instruct
+ppo_mini_batch_size:
+    value: 32
+ray_kwargs:
+    value:
+        ray_init:
+            num_cpus: null
+        timeline_json_file: null
+reward_model:
+    value:
+        enable: false
+        enable_resource_pool: false
+        forward_max_token_len_per_gpu: 32768
+        launch_reward_fn_async: false
+        max_length: null
+        micro_batch_size: null
+        micro_batch_size_per_gpu: null
+        model:
+            external_lib: null
+            fsdp_config:
+                _target_: verl.workers.config.FSDPEngineConfig
+                forward_prefetch: false
+                fsdp_size: -1
+                param_offload: false
+                reshard_after_forward: true
+                wrap_policy:
+                    min_num_params: 0
+            input_tokenizer: Qwen/Qwen2.5-0.5B-Instruct
+            path: ~/models/FsfairX-LLaMA3-RM-v0.1
+            trust_remote_code: false
+            use_fused_kernels: false
+            use_remove_padding: false
+            use_shm: false
+        n_gpus_per_node: 0
+        nnodes: 0
+        profiler:
+            _target_: verl.utils.profiler.ProfilerConfig
+            all_ranks: false
+            enable: false
+            ranks: []
+            save_path: outputs/profile
+            tool: null
+            tool_config:
+                npu:
+                    _target_: verl.utils.profiler.config.NPUToolConfig
+                    analysis: true
+                    contents: []
+                    discrete: false
+                    level: level1
+                nsys:
+                    _target_: verl.utils.profiler.config.NsightToolConfig
+                    discrete: false
+                torch:
+                    _target_: verl.utils.profiler.config.TorchProfilerToolConfig
+                    step_end: null
+                    step_start: 0
+                torch_memory:
+                    _target_: verl.utils.profiler.config.TorchMemoryToolConfig
+                    stack_depth: 32
+                    trace_alloc_max_entries: 100000
+        reward_manager: naive
+        sandbox_fusion:
+            max_concurrent: 64
+            memory_limit_mb: 1024
+            url: null
+        strategy: fsdp
+        ulysses_sequence_parallel_size: 1
+        use_dynamic_bsz: false
+seed:
+    value:
+        train: 10000
+        val: 123
+system:
+    value:
+        CUDA_VISIBLE_DEVICES: 0,1,2,3,4,5,6,7
+trainer:
+    value:
+        balance_batch: true
+        critic_warmup: 0
+        default_hdfs_dir: null
+        default_local_dir: /mnt/general/wanghy/RAGEN/saves/
+        del_local_ckpt_after_load: false
+        device: cuda
+        esi_redundant_time: 0
+        experiment_name: rubikscube-1-starpos-1218
+        generations_to_log_to_wandb:
+            train: 128
+            val: 20
+        local_log_dir: results/
+        log_val_generations: 0
+        logger:
+            - console
+            - wandb
+        max_actor_ckpt_to_keep: 1
+        max_critic_ckpt_to_keep: 1
+        n_gpus_per_node: 8
+        nnodes: 1
+        project_name: ragen_latest_qwen_05B_it
+        ray_wait_register_center_timeout: 300
+        resume_from_path: null
+        resume_mode: auto
+        rollout_data_dir: null
+        save_freq: -1
+        test_freq: 10
+        total_epochs: 30
+        total_training_steps: 200
+        use_legacy_worker_impl: auto
+        val_before_train: true
+        val_only: false
+        validation_data_dir: null
+        validation_steps: 1

wandb/run-20251218_190243-6eiq1qnz/files/media/table/val/generations_0_6cb7f39cae32c2b9f416.table.json

@@ -0,0 +1 @@
+{"columns": ["step", "input_1", "output_1", "score_1", "input_2", "output_2", "score_2", "input_3", "output_3", "score_3", "input_4", "output_4", "score_4", "input_5", "output_5", "score_5", "input_6", "output_6", "score_6", "input_7", "output_7", "score_7", "input_8", "output_8", "score_8", "input_9", "output_9", "score_9", "input_10", "output_10", "score_10", "input_11", "output_11", "score_11", "input_12", "output_12", "score_12", "input_13", "output_13", "score_13", "input_14", "output_14", "score_14", "input_15", "output_15", "score_15", "input_16", "output_16", "score_16", "input_17", "output_17", "score_17", "input_18", "output_18", "score_18", "input_19", "output_19", "score_19", "input_20", "output_20", "score_20"], "data": [[0, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, B]\n [W, B]\nLeft (L):  [O, O]\n [O, O]\nFront (F): [G, W]\n [G, W]\nRight (R): [R, R]\n [R, R]\nBack (B):  [Y, B]\n [Y, B]\nDown (D):  [Y, G]\n [Y, G]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think> [You are about to make a move that will reverse the order of the colors on the cube.] </think> <answer> [U' || D' || L' || R' || F' || B' || L' || R' || U']\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, B]\n [W, B]\nLeft (L):  [O, O]\n [O, O]\nFront (F): [G, W]\n [G, W]\nRight (R): [R, R]\n [R, R]\nBack (B):  [Y, B]\n [Y, B]\nDown (D):  [Y, G]\n [Y, G]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think> [You are about to make a move that will reverse the order of the colors on the cube.] </think> <answer> [U' || D' || L' || R' || F' || B' || L' || R' || U']\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, B]\n [W, B]\nLeft (L):  [O, O]\n [O, O]\nFront (F): [G, W]\n [G, W]\nRight (R): [R, R]\n [R, R]\nBack (B):  [Y, B]\n [Y, B]\nDown (D):  [Y, G]\n [Y, G]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think> [You are about to make a move that will reverse the order of the colors on the cube.] </think> <answer> [U' || D' || L' || R' || F' || B' || L' || R' || U']\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, B]\n [W, B]\nLeft (L):  [O, O]\n [O, O]\nFront (F): [G, W]\n [G, W]\nRight (R): [R, R]\n [R, R]\nBack (B):  [Y, B]\n [Y, B]\nDown (D):  [Y, G]\n [Y, G]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think> [You are about to make a move that will reverse the order of the colors on the cube.] </think> <answer> [U' || D' || L' || R' || F' || B' || L' || R' || U']\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, B]\n [W, B]\nLeft (L):  [O, O]\n [O, O]\nFront (F): [G, W]\n [G, W]\nRight (R): [R, R]\n [R, R]\nBack (B):  [Y, B]\n [Y, B]\nDown (D):  [Y, G]\n [Y, G]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think> [You are about to make a move that will reverse the order of the colors on the cube.] </think> <answer> [U' || D' || L' || R' || F' || B' || L' || R' || U']\n", -2.5, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 15/20\n\nUp (U):    [O, W]\n [G, B]\nLeft (L):  [G, Y]\n [G, Y]\nFront (F): [R, R]\n [O, O]\nRight (R): [Y, B]\n [R, B]\nBack (B):  [O, R]\n [G, W]\nDown (D):  [W, W]\n [B, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 5 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Perform a 90\u00b0 clockwise rotation on the right face again.</think><answer>R</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 16/20\n\nUp (U):    [O, R]\n [G, O]\nLeft (L):  [G, Y]\n [G, Y]\nFront (F): [R, W]\n [O, Y]\nRight (R): [R, Y]\n [B, B]\nBack (B):  [B, R]\n [W, W]\nDown (D):  [W, G]\n [B, O]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 4 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Perform a 90\u00b0 clockwise rotation on the right face again.</think><answer>R</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 17/20\n\nUp (U):    [O, W]\n [G, Y]\nLeft (L):  [G, Y]\n [G, Y]\nFront (F): [R, G]\n [O, O]\nRight (R): [B, R]\n [B, Y]\nBack (B):  [O, R]\n [R, W]\nDown (D):  [W, W]\n [B, B]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 3 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Perform a 90\u00b0 clockwise rotation on the front face again.</think><answer>F</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 18/20\n\nUp (U):    [O, W]\n [Y, Y]\nLeft (L):  [G, W]\n [G, W]\nFront (F): [O, R]\n [O, G]\nRight (R): [G, R]\n [Y, Y]\nBack (B):  [O, R]\n [R, W]\nDown (D):  [B, B]\n [B, B]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 2 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Perform a 90\u00b0 clockwise rotation on the left face again.</think><answer>L</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 19/20\n\nUp (U):    [W, W]\n [R, Y]\nLeft (L):  [G, G]\n [W, W]\nFront (F): [O, R]\n [Y, G]\nRight (R): [G, R]\n [Y, Y]\nBack (B):  [O, B]\n [R, B]\nDown (D):  [O, B]\n [O, B]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 1 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Perform a 90\u00b0 clockwise rotation on the back face again.</think><answer>B</answer>\n", 0.0, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [B, B]\nFront (F): [G, G]\n [O, O]\nRight (R): [R, R]\n [G, G]\nBack (B):  [B, B]\n [R, R]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>U, U', D, D', L, L', R, R', F, F', B, B'</thep answer> U</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [B, B]\nFront (F): [G, G]\n [O, O]\nRight (R): [R, R]\n [G, G]\nBack (B):  [B, B]\n [R, R]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>U, U', D, D', L, L', R, R', F, F', B, B'</thep answer> U</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [B, B]\nFront (F): [G, G]\n [O, O]\nRight (R): [R, R]\n [G, G]\nBack (B):  [B, B]\n [R, R]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>U, U', D, D', L, L', R, R', F, F', B, B'</thep answer> U</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [B, B]\nFront (F): [G, G]\n [O, O]\nRight (R): [R, R]\n [G, G]\nBack (B):  [B, B]\n [R, R]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>U, U', D, D', L, L', R, R', F, F', B, B'</thep answer> U</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [B, B]\nFront (F): [G, G]\n [O, O]\nRight (R): [R, R]\n [G, G]\nBack (B):  [B, B]\n [R, R]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>U, U', D, D', L, L', R, R', F, F', B, B'</thep answer> U</answer>\n", -2.5, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [O, O]\nLeft (L):  [O, Y]\n [O, Y]\nFront (F): [G, G]\n [G, G]\nRight (R): [W, R]\n [W, R]\nBack (B):  [B, B]\n [B, B]\nDown (D):  [R, R]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>After performing all the steps, I have completed the puzzle.</think><answer>Congratulations! You have successfully solved the Rubik's Cube.</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [O, O]\nLeft (L):  [O, Y]\n [O, Y]\nFront (F): [G, G]\n [G, G]\nRight (R): [W, R]\n [W, R]\nBack (B):  [B, B]\n [B, B]\nDown (D):  [R, R]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>After performing all the steps, I have completed the puzzle.</think><answer>Congratulations! You have successfully solved the Rubik's Cube.</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [O, O]\nLeft (L):  [O, Y]\n [O, Y]\nFront (F): [G, G]\n [G, G]\nRight (R): [W, R]\n [W, R]\nBack (B):  [B, B]\n [B, B]\nDown (D):  [R, R]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>After performing all the steps, I have completed the puzzle.</think><answer>Congratulations! You have successfully solved the Rubik's Cube.</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [O, O]\nLeft (L):  [O, Y]\n [O, Y]\nFront (F): [G, G]\n [G, G]\nRight (R): [W, R]\n [W, R]\nBack (B):  [B, B]\n [B, B]\nDown (D):  [R, R]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>After performing all the steps, I have completed the puzzle.</think><answer>Congratulations! You have successfully solved the Rubik's Cube.</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [O, O]\nLeft (L):  [O, Y]\n [O, Y]\nFront (F): [G, G]\n [G, G]\nRight (R): [W, R]\n [W, R]\nBack (B):  [B, B]\n [B, B]\nDown (D):  [R, R]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>After performing all the steps, I have completed the puzzle.</think><answer>Congratulations! You have successfully solved the Rubik's Cube.</answer>\n", -2.5, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [R, R]\n [W, W]\nLeft (L):  [W, O]\n [W, O]\nFront (F): [G, G]\n [G, G]\nRight (R): [R, Y]\n [R, Y]\nBack (B):  [B, B]\n [B, B]\nDown (D):  [Y, Y]\n [O, O]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[You are about to perform a 90-degree clockwise rotation on the front face.]</think><answer>[Perform a 90-degree clockwise rotation on the front face.]</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [R, R]\n [W, W]\nLeft (L):  [W, O]\n [W, O]\nFront (F): [G, G]\n [G, G]\nRight (R): [R, Y]\n [R, Y]\nBack (B):  [B, B]\n [B, B]\nDown (D):  [Y, Y]\n [O, O]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[You are about to perform a 90-degree counter-clockwise rotation on the front face.]</think><answer>Perform a 90-degree counter-clockwise rotation on the front face.</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [R, R]\n [W, W]\nLeft (L):  [W, O]\n [W, O]\nFront (F): [G, G]\n [G, G]\nRight (R): [R, Y]\n [R, Y]\nBack (B):  [B, B]\n [B, B]\nDown (D):  [Y, Y]\n [O, O]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[You have completed the 2x2 Rubik's Cube.]</think><answer>Congratulations! You have successfully restored the 2x2 Rubik's Cube to a single, unique color.</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [R, R]\n [W, W]\nLeft (L):  [W, O]\n [W, O]\nFront (F): [G, G]\n [G, G]\nRight (R): [R, Y]\n [R, Y]\nBack (B):  [B, B]\n [B, B]\nDown (D):  [Y, Y]\n [O, O]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[You have completed the 2x2 Rubik's Cube.]</think><answer>Congratulations! You have successfully restored the 2x2 Rubik's Cube to a single, unique color.</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [R, R]\n [W, W]\nLeft (L):  [W, O]\n [W, O]\nFront (F): [G, G]\n [G, G]\nRight (R): [R, Y]\n [R, Y]\nBack (B):  [B, B]\n [B, B]\nDown (D):  [Y, Y]\n [O, O]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[You have completed the 2x2 Rubik's Cube.]</think><answer>Congratulations! You have successfully restored the 2x2 Rubik's Cube to a single, unique color.</answer>\n", -2.5, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 15/20\n\nUp (U):    [B, R]\n [W, R]\nLeft (L):  [Y, B]\n [Y, B]\nFront (F): [O, B]\n [O, W]\nRight (R): [W, W]\n [G, G]\nBack (B):  [G, R]\n [Y, R]\nDown (D):  [Y, O]\n [G, O]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 5 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Make 2 face rotations to restore the bottom face.</think><answer>R</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 16/20\n\nUp (U):    [B, B]\n [W, W]\nLeft (L):  [Y, B]\n [Y, B]\nFront (F): [O, O]\n [O, O]\nRight (R): [G, W]\n [G, W]\nBack (B):  [R, R]\n [R, R]\nDown (D):  [Y, Y]\n [G, G]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 4 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Make 2 face rotations to restore the bottom face.</think><answer>R</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 17/20\n\nUp (U):    [B, O]\n [W, O]\nLeft (L):  [Y, B]\n [Y, B]\nFront (F): [O, Y]\n [O, G]\nRight (R): [G, G]\n [W, W]\nBack (B):  [W, R]\n [B, R]\nDown (D):  [Y, R]\n [G, R]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 3 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Make 2 face rotations to restore the bottom face.</think><answer>R</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 18/20\n\nUp (U):    [B, Y]\n [W, G]\nLeft (L):  [Y, B]\n [Y, B]\nFront (F): [O, R]\n [O, R]\nRight (R): [W, G]\n [W, G]\nBack (B):  [O, R]\n [O, R]\nDown (D):  [Y, B]\n [G, W]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 2 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Make 2 face rotations to restore the bottom face.</think><answer>R</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 19/20\n\nUp (U):    [B, R]\n [W, R]\nLeft (L):  [Y, B]\n [Y, B]\nFront (F): [O, B]\n [O, W]\nRight (R): [W, W]\n [G, G]\nBack (B):  [G, R]\n [Y, R]\nDown (D):  [Y, O]\n [G, O]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 1 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Make 2 face rotations to restore the bottom face.</think><answer>R</answer>\n", 0.0, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[This is the initial state of the Rubik's Cube.]</think><answer>[U]</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[The cube is already solved.]</think><answer>[No move needed.]</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[The cube is already solved.]</think><answer>[No move needed.]</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[This is the initial state of the Rubik's Cube.]</think><answer>[No move needed.]</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>U</think><answer>U</answer>\n", 0.6000000238418579, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [G, G]\nFront (F): [G, G]\n [R, R]\nRight (R): [R, R]\n [B, B]\nBack (B):  [B, B]\n [O, O]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Rotate the center face clockwise to U''</think><answer>U''</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [G, G]\nFront (F): [G, G]\n [R, R]\nRight (R): [R, R]\n [B, B]\nBack (B):  [B, B]\n [O, O]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Rotate the center face clockwise to U''</think><answer>U''</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [G, G]\nFront (F): [G, G]\n [R, R]\nRight (R): [R, R]\n [B, B]\nBack (B):  [B, B]\n [O, O]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Rotate the center face clockwise to U''</think><answer>U''</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [G, G]\nFront (F): [G, G]\n [R, R]\nRight (R): [R, R]\n [B, B]\nBack (B):  [B, B]\n [O, O]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Rotate the center face clockwise to U''</think><answer>U''</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [G, G]\nFront (F): [G, G]\n [R, R]\nRight (R): [R, R]\n [B, B]\nBack (B):  [B, B]\n [O, O]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Rotate the center face clockwise to U''</think><answer>U''</answer>\n", -2.5, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 15/20\n\nUp (U):    [R, W]\n [R, W]\nLeft (L):  [B, B]\n [W, O]\nFront (F): [W, O]\n [G, G]\nRight (R): [G, G]\n [R, Y]\nBack (B):  [R, Y]\n [B, B]\nDown (D):  [Y, Y]\n [O, O]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 5 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Aligning the top face with the bottom face will help us solve the cube.</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 16/20\n\nUp (U):    [R, R]\n [W, W]\nLeft (L):  [W, O]\n [W, O]\nFront (F): [G, G]\n [G, G]\nRight (R): [R, Y]\n [R, Y]\nBack (B):  [B, B]\n [B, B]\nDown (D):  [Y, Y]\n [O, O]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 4 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Aligning the top face with the bottom face will help us solve the cube.</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 17/20\n\nUp (U):    [W, R]\n [W, R]\nLeft (L):  [G, G]\n [W, O]\nFront (F): [R, Y]\n [G, G]\nRight (R): [B, B]\n [R, Y]\nBack (B):  [W, O]\n [B, B]\nDown (D):  [Y, Y]\n [O, O]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 3 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Aligning the top face with the bottom face will help us solve the cube.</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 18/20\n\nUp (U):    [W, W]\n [R, R]\nLeft (L):  [R, Y]\n [W, O]\nFront (F): [B, B]\n [G, G]\nRight (R): [W, O]\n [R, Y]\nBack (B):  [G, G]\n [B, B]\nDown (D):  [Y, Y]\n [O, O]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 2 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Aligning the top face with the bottom face will help us solve the cube.</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 19/20\n\nUp (U):    [R, W]\n [R, W]\nLeft (L):  [B, B]\n [W, O]\nFront (F): [W, O]\n [G, G]\nRight (R): [G, G]\n [R, Y]\nBack (B):  [R, Y]\n [B, B]\nDown (D):  [Y, Y]\n [O, O]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 1 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Aligning the top face with the bottom face will help us solve the cube.</think><answer>U</answer>\n", 0.0, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 13/20\n\nUp (U):    [O, W]\n [O, W]\nLeft (L):  [G, G]\n [O, Y]\nFront (F): [W, R]\n [G, G]\nRight (R): [B, B]\n [W, R]\nBack (B):  [O, Y]\n [B, B]\nDown (D):  [R, R]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 7 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>U</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 14/20\n\nUp (U):    [O, O]\n [W, W]\nLeft (L):  [W, R]\n [O, Y]\nFront (F): [B, B]\n [G, G]\nRight (R): [O, Y]\n [W, R]\nBack (B):  [G, G]\n [B, B]\nDown (D):  [R, R]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 6 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>U</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 15/20\n\nUp (U):    [W, O]\n [W, O]\nLeft (L):  [B, B]\n [O, Y]\nFront (F): [O, Y]\n [G, G]\nRight (R): [G, G]\n [W, R]\nBack (B):  [W, R]\n [B, B]\nDown (D):  [R, R]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 5 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>U</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 16/20\n\nUp (U):    [W, W]\n [O, O]\nLeft (L):  [O, Y]\n [O, Y]\nFront (F): [G, G]\n [G, G]\nRight (R): [W, R]\n [W, R]\nBack (B):  [B, B]\n [B, B]\nDown (D):  [R, R]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 4 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>U</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 17/20\n\nUp (U):    [O, W]\n [O, W]\nLeft (L):  [G, G]\n [O, Y]\nFront (F): [W, R]\n [G, G]\nRight (R): [B, B]\n [W, R]\nBack (B):  [O, Y]\n [B, B]\nDown (D):  [R, R]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 3 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>U</think><answer>U</answer>\n", -0.699999988079071, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 15/20\n\nUp (U):    [B, W]\n [B, W]\nLeft (L):  [O, O]\n [B, Y]\nFront (F): [W, G]\n [O, O]\nRight (R): [R, R]\n [W, G]\nBack (B):  [B, Y]\n [R, R]\nDown (D):  [G, G]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 5 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Now I need to solve the bottom layer again. I can use U to move the bottom layer to the top layer.</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 16/20\n\nUp (U):    [B, B]\n [W, W]\nLeft (L):  [W, G]\n [B, Y]\nFront (F): [R, R]\n [O, O]\nRight (R): [B, Y]\n [W, G]\nBack (B):  [O, O]\n [R, R]\nDown (D):  [G, G]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 4 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Now I need to solve the bottom layer again. I can use U to move the bottom layer to the top layer.</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 17/20\n\nUp (U):    [W, B]\n [W, B]\nLeft (L):  [R, R]\n [B, Y]\nFront (F): [B, Y]\n [O, O]\nRight (R): [O, O]\n [W, G]\nBack (B):  [W, G]\n [R, R]\nDown (D):  [G, G]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 3 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Now I need to solve the bottom layer again. I can use U to move the bottom layer to the top layer.</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 18/20\n\nUp (U):    [W, W]\n [B, B]\nLeft (L):  [B, Y]\n [B, Y]\nFront (F): [O, O]\n [O, O]\nRight (R): [W, G]\n [W, G]\nBack (B):  [R, R]\n [R, R]\nDown (D):  [G, G]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 2 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Now I need to solve the bottom layer again. I can use U to move the bottom layer to the top layer.</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 19/20\n\nUp (U):    [B, W]\n [B, W]\nLeft (L):  [O, O]\n [B, Y]\nFront (F): [W, G]\n [O, O]\nRight (R): [R, R]\n [W, G]\nBack (B):  [B, Y]\n [R, R]\nDown (D):  [G, G]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 1 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Now I need to solve the bottom layer again. I can use U to move the bottom layer to the top layer.</think><answer>U</answer>\n", 0.0, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [G, G]\nFront (F): [G, G]\n [R, R]\nRight (R): [R, R]\n [B, B]\nBack (B):  [B, B]\n [O, O]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[You have completed the 2x2 Rubik's Cube.]</think><answer>[Congratulations! You have solved the 2x2 Rubik's Cube.]</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [G, G]\nFront (F): [G, G]\n [R, R]\nRight (R): [R, R]\n [B, B]\nBack (B):  [B, B]\n [O, O]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[You have completed the 2x2 Rubik's Cube.]</think><answer>[Congratulations! You have solved the 2x2 Rubik's Cube.]</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [G, G]\nFront (F): [G, G]\n [R, R]\nRight (R): [R, R]\n [B, B]\nBack (B):  [B, B]\n [O, O]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[You have completed the 2x2 Rubik's Cube.]</think><answer>[Congratulations! You have solved the 2x2 Rubik's Cube.]</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [G, G]\nFront (F): [G, G]\n [R, R]\nRight (R): [R, R]\n [B, B]\nBack (B):  [B, B]\n [O, O]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[You have completed the 2x2 Rubik's Cube.]</think><answer>[Congratulations! You have solved the 2x2 Rubik's Cube.]</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [G, G]\nFront (F): [G, G]\n [R, R]\nRight (R): [R, R]\n [B, B]\nBack (B):  [B, B]\n [O, O]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[You have completed the 2x2 Rubik's Cube.]</think><answer>[Congratulations! You have solved the 2x2 Rubik's Cube.]</answer>\n", -2.5, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 4/20\n\nUp (U):    [R, R]\n [G, G]\nLeft (L):  [G, O]\n [G, O]\nFront (F): [Y, W]\n [Y, W]\nRight (R): [O, B]\n [O, B]\nBack (B):  [W, W]\n [Y, Y]\nDown (D):  [B, B]\n [R, R]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 16 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Perform a 90\u00b0 counter-clockwise rotation on the right face again.</think><answer>R\"</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 4/20\n\nUp (U):    [R, R]\n [G, G]\nLeft (L):  [G, O]\n [G, O]\nFront (F): [Y, W]\n [Y, W]\nRight (R): [O, B]\n [O, B]\nBack (B):  [W, W]\n [Y, Y]\nDown (D):  [B, B]\n [R, R]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 16 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Perform a 90\u00b0 counter-clockwise rotation on the right face again.</think><answer>R\"</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 4/20\n\nUp (U):    [R, R]\n [G, G]\nLeft (L):  [G, O]\n [G, O]\nFront (F): [Y, W]\n [Y, W]\nRight (R): [O, B]\n [O, B]\nBack (B):  [W, W]\n [Y, Y]\nDown (D):  [B, B]\n [R, R]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 16 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Perform a 90\u00b0 counter-clockwise rotation on the right face again.</think><answer>R\"</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 4/20\n\nUp (U):    [R, R]\n [G, G]\nLeft (L):  [G, O]\n [G, O]\nFront (F): [Y, W]\n [Y, W]\nRight (R): [O, B]\n [O, B]\nBack (B):  [W, W]\n [Y, Y]\nDown (D):  [B, B]\n [R, R]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 16 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Perform a 90\u00b0 counter-clockwise rotation on the right face again.</think><answer>R\"</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 4/20\n\nUp (U):    [R, R]\n [G, G]\nLeft (L):  [G, O]\n [G, O]\nFront (F): [Y, W]\n [Y, W]\nRight (R): [O, B]\n [O, B]\nBack (B):  [W, W]\n [Y, Y]\nDown (D):  [B, B]\n [R, R]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 16 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Perform a 90\u00b0 counter-clockwise rotation on the right face again.</think><answer>R\"</answer>\n", -2.0999999046325684, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [G, G]\n [O, O]\nFront (F): [R, R]\n [G, G]\nRight (R): [B, B]\n [R, R]\nBack (B):  [O, O]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Make a face rotation to align the middle and left faces again.</think><answer>U, U', D, D', L, L', R, R', F, F', B, B'</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [G, G]\n [O, O]\nFront (F): [R, R]\n [G, G]\nRight (R): [B, B]\n [R, R]\nBack (B):  [O, O]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Make a face rotation to align the middle and right faces again.</think><answer>U, U', D, D', L, L', R, R', F, F', B, B'</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [G, G]\n [O, O]\nFront (F): [R, R]\n [G, G]\nRight (R): [B, B]\n [R, R]\nBack (B):  [O, O]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Make a face rotation to align the middle and top faces again.</think><answer>U, U', D, D', L, L', R, R', F, F', B, B'</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [G, G]\n [O, O]\nFront (F): [R, R]\n [G, G]\nRight (R): [B, B]\n [R, R]\nBack (B):  [O, O]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Make a face rotation to align the middle and bottom faces again.</think><answer>U, U', D, D', L, L', R, R', F, F', B, B'</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [G, G]\n [O, O]\nFront (F): [R, R]\n [G, G]\nRight (R): [B, B]\n [R, R]\nBack (B):  [O, O]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Make a face rotation to align the middle and left faces again.</think><answer>U, U', D, D', L, L', R, R', F, F', B, B'</answer>\n", -2.5, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [B, B]\nFront (F): [G, G]\n [O, O]\nRight (R): [R, R]\n [G, G]\nBack (B):  [B, B]\n [R, R]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[The cube is already in the correct orientation.]</think><answer>[No action needed.]</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [B, B]\nFront (F): [G, G]\n [O, O]\nRight (R): [R, R]\n [G, G]\nBack (B):  [B, B]\n [R, R]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[The cube is already in the correct orientation.]</think><answer>[No action needed.]</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [O, O]\n [B, B]\nFront (F): [G, G]\n [O, O]\nRight (R): [R, R]\n [G, G]\nBack (B):  [B, B]\n [R, R]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>U</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 1/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [G, G]\n [B, B]\nFront (F): [R, R]\n [O, O]\nRight (R): [B, B]\n [G, G]\nBack (B):  [O, O]\n [R, R]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 19 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>U</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 2/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [R, R]\n [B, B]\nFront (F): [B, B]\n [O, O]\nRight (R): [O, O]\n [G, G]\nBack (B):  [G, G]\n [R, R]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 18 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>U</think><answer>U</answer>\n", 0.800000011920929, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[You are about to make a move that will reverse the order of the colors on the cube.]</think><answer>[U]</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[You are about to make a move that will reverse the order of the colors on the cube.]</think><answer>[U]</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[You are about to reverse the order of the colors on the cube.]</think><answer>[U]</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>[U]</think><answer>[U]</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>U</think><answer>U</answer>\n", 0.10000002384185791, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Rotate the top left face clockwise to U''</think><answer>U''</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Rotate the top right face clockwise to U''</think><answer>U''</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Rotate the bottom right face clockwise to U''</think><answer>U''</answer>\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Rotate the bottom left face clockwise to U''</think><answer>U''</answer>\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Rotate the top face clockwise to U</think><answer>U</answer>\n", -0.10000002384185791, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, B]\n [W, B]\nLeft (L):  [O, O]\n [O, O]\nFront (F): [G, W]\n [G, W]\nRight (R): [R, R]\n [R, R]\nBack (B):  [Y, B]\n [Y, B]\nDown (D):  [Y, G]\n [Y, G]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think> [You are about to make a move that will disrupt the normal order of the cube. This move will cause the cube to rotate and create a pattern that will make it impossible to restore the cube. You need to find a way to restore the cube while keeping the faces identical. Consider using a clockwise rotation of the cube to achieve this. ] </think> <answer> [U, U', D, D', L, L', R, R', F, F', B, B']\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, B]\n [W, B]\nLeft (L):  [O, O]\n [O, O]\nFront (F): [G, W]\n [G, W]\nRight (R): [R, R]\n [R, R]\nBack (B):  [Y, B]\n [Y, B]\nDown (D):  [Y, G]\n [Y, G]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think> [You are about to make a move that will disrupt the normal order of the cube. This move will cause the cube to rotate and create a pattern that will make it impossible to restore the cube. You need to find a way to restore the cube while keeping the faces identical. Consider using a clockwise rotation of the cube to achieve this. ] </think> <answer> [U, U', D, D', L, L', R, R', F, F', B, B']\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, B]\n [W, B]\nLeft (L):  [O, O]\n [O, O]\nFront (F): [G, W]\n [G, W]\nRight (R): [R, R]\n [R, R]\nBack (B):  [Y, B]\n [Y, B]\nDown (D):  [Y, G]\n [Y, G]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think> [You are about to make a move that will disrupt the normal order of the cube. This move will cause the cube to rotate and create a pattern that will make it impossible to restore the cube. You need to find a way to restore the cube while keeping the faces identical. Consider using a clockwise rotation of the cube to achieve this. ] </think> <answer> [U, U', D, D', L, L', R, R', F, F', B, B']\nuser\nReward:\n0.0\n\nTurn 4:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, B]\n [W, B]\nLeft (L):  [O, O]\n [O, O]\nFront (F): [G, W]\n [G, W]\nRight (R): [R, R]\n [R, R]\nBack (B):  [Y, B]\n [Y, B]\nDown (D):  [Y, G]\n [Y, G]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think> [You are about to make a move that will disrupt the normal order of the cube. This move will cause the cube to rotate and create a pattern that will make it impossible to restore the cube. You need to find a way to restore the cube while keeping the faces identical. Consider using a clockwise rotation of the cube to achieve this. ] </think> <answer> [U, U', D, D', L, L', R, R', F, F', B, B']\nuser\nReward:\n0.0\n\nTurn 5:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, B]\n [W, B]\nLeft (L):  [O, O]\n [O, O]\nFront (F): [G, W]\n [G, W]\nRight (R): [R, R]\n [R, R]\nBack (B):  [Y, B]\n [Y, B]\nDown (D):  [Y, G]\n [Y, G]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think> [You are about to make a move that will disrupt the normal order of the cube. This move will cause the cube to rotate and create a pattern that will make it impossible to restore the cube. You need to find a way to restore the cube while keeping the faces identical. Consider using a clockwise rotation of the cube to achieve this. ] </think> <answer> [U, U', D, D', L, L', R, R', F, F', B, B']\n", -2.5, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>After U, U', D, D', L, L', R, R', F, F', B, B', the cube is upside down and faces are reversed. This is the first move.</think><answer>U</answer>\n", 1.0, "", "system\nYou're a helpful assistant. \nuser\nYou are solving a 2x2 Rubik's Cube (Pocket Cube). The goal is to restore the cube so that each of the faces consists of a single, unique color.\nAvailable actions use standard Singmaster notation for face rotations: U, U', D, D', L, L', R, R', F, F', B, B'.\n- Faces: U (Up), D (Down), L (Left), R (Right), F (Front), B (Back).\n- Modifiers: A letter alone means 90\u00b0 clockwise (e.g., 'R'). A letter with prime (') means 90\u00b0 counter-clockwise (e.g., \"R'\").\nRespond with a sequence of actions separated by \"||\".\nExample: <answer>U</answer>\n\nYour available actions are:\nU, U', D, D', L, L', R, R', F, F', B, B'\nYou can make up to 20 actions, separated by the action separator \" || \"\n\nTurn 1:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 0/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [G, G]\n [O, O]\nFront (F): [R, R]\n [G, G]\nRight (R): [B, B]\n [R, R]\nBack (B):  [O, O]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 20 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>You can use the following actions to solve the Rubik's Cube: U, U', D, D', L, L', R, R', F, F', B, B'</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 2:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 1/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [R, R]\n [O, O]\nFront (F): [B, B]\n [G, G]\nRight (R): [O, O]\n [R, R]\nBack (B):  [G, G]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 19 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Using the available actions: U, U', D, D', L, L', R, R', F, F', B, B'</think><answer>U</answer>\nuser\nReward:\n0.0\n\nTurn 3:\nState:\n=== Rubik's Cube 2x2 State ===\nStep: 2/20\n\nUp (U):    [W, W]\n [W, W]\nLeft (L):  [B, B]\n [O, O]\nFront (F): [O, O]\n [G, G]\nRight (R): [G, G]\n [R, R]\nBack (B):  [R, R]\n [B, B]\nDown (D):  [Y, Y]\n [Y, Y]\n\nAvailable Actions:\nU, U', D, D', L, L', R, R', F, F', B, B'\n\nWhat is your next move?\nYou have 18 actions left. Always output: <think> [Your thoughts] </think> <answer> [your answer] </answer> with no extra text. Strictly follow this format. Max response length: 96 words (tokens).\n\nassistant\n<think>Using the available actions: U, U', D, D', L, L', R, R', F, F', B, B'</think><answer>U</answer>\n", 1.0]]}