set -x # Colocated GRPO training+generation for Qwen3-8B on the SWE-Bench task. # Uses 1 node with 8 GPUs. # uv run --isolated examples/train/mini_swe_agent/preprocess_swegym.py --output_dir ~/data/swe_gym_subset # bash examples/train/mini_swe_agent/run_mini_swe_8B.sh DATA_DIR="$HOME/data/swe_gym_subset" CKPT_PATH="$HOME/ckpts/llm_mini_swe" # Save trajectories here for debugging # NOTE: For a multi-node cluster, ensure that this is on NFS so that you can save all trajectories in the same path MINISWE_TRAJ_DIR="$HOME/mini_swe_agent_trajs" NUM_GPUS=8 NNODES=1 NUM_INFERENCE_ENGINES=4 TP_SIZE=2 LOGGER=wandb # We use a small batch size here for demonstration # NOTE (sumanthrh): The `generator.max_turns` here is actually unused, and we use the `step_limit` from the `swebench.yaml` file. # This simply has to be a value > 1 uv run --isolated --extra fsdp --extra miniswe --env-file examples/train/mini_swe_agent/.env.miniswe -m examples.train.mini_swe_agent.main_mini_swe \ data.train_data="['$DATA_DIR/train.parquet']" \ data.val_data="['$DATA_DIR/validation.parquet']" \ trainer.algorithm.advantage_estimator="grpo" \ trainer.policy.model.path="Qwen/Qwen3-8B" \ trainer.placement.colocate_all=true \ trainer.strategy=fsdp \ trainer.placement.policy_num_gpus_per_node=$NUM_GPUS \ trainer.placement.ref_num_gpus_per_node=$NUM_GPUS \ trainer.placement.policy_num_nodes=$NNODES \ trainer.placement.ref_num_nodes=$NNODES \ trainer.policy.sequence_parallel_size=2 \ generator.inference_engine.num_engines=$NUM_INFERENCE_ENGINES \ generator.inference_engine.tensor_parallel_size=$TP_SIZE \ trainer.epochs=20 \ trainer.eval_batch_size=50 \ trainer.eval_before_train=true \ trainer.eval_interval=5 \ trainer.update_epochs_per_batch=1 \ trainer.train_batch_size=16 \ trainer.policy_mini_batch_size=16 \ trainer.micro_forward_batch_size_per_gpu=1 \ trainer.micro_train_batch_size_per_gpu=1 \ trainer.dump_data_batch=true \ trainer.ckpt_interval=10 \ trainer.max_prompt_length=4096 \ generator.sampling_params.max_generate_length=4096 \ generator.max_input_length=30720 \ generator.max_turns=20 \ trainer.policy.optimizer_config.lr=1.0e-6 \ trainer.algorithm.use_kl_loss=true \ generator.inference_engine.backend=vllm \ generator.inference_engine.run_engines_locally=True \ generator.inference_engine.weight_sync_backend=nccl \ generator.batched=true \ generator.n_samples_per_prompt=4 \ generator.inference_engine.gpu_memory_utilization=0.8 \ trainer.logger="$LOGGER" \ trainer.project_name="mini_swe" \ trainer.run_name="mini_swe_8B_swe_gym" \ trainer.resume_mode=null \ trainer.ckpt_path="$CKPT_PATH" \ generator.miniswe_config_path="examples/mini_swe_agent/swebench.yaml" \ generator.miniswe_traj_dir=$MINISWE_TRAJ_DIR $@