MTSQL-R1/run_text2sql.sh at main · taichengguo/MTSQL-R1 · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
set -x

ulimit -n 65535
data="cosql"
#data="sparc"
DATA_FOLDER="data/$data"
#DATA_FOLDER="PATH/TO/data/gsm8k/"

# Dataset: Except Simple
# Except correct
PROJECT_DIR="$(pwd)"
CONFIG_PATH="$PROJECT_DIR/examples/sglang_multiturn/config"

#MODEL_PATH="Qwen/Qwen3-0.6B"
#MODEL_PATH="Qwen/Qwen3-1.7B"

#clip_ratio_low=0.2
#clip_ratio_high=0.28

# Prev - best model
#tag="4B"
tag="17B"
#MODEL_PATH="../llamafactory/saves/sql/${data}_round3_${tag}/checkpoint-2054"  # 4B
MODEL_PATH="../llamafactory/saves/sql/${data}_round3_${tag}/checkpoint-4108"  #17B


#ray job submit --address="http://127.0.0.1:8265" \
#    --runtime-env=verl/trainer/runtime_env.yaml \
#    --no-wait \
#    -- \
python3 -m verl.trainer.main_ppo \
    --config-path="$CONFIG_PATH" \
    --config-name='text2sql_multiturn_grpo' \
    custom_reward_function.path=verl/utils/reward_score/text2sql.py  \
    algorithm.adv_estimator=grpo \
    data.train_files=$DATA_FOLDER/train_sft.parquet \
    data.val_files=$DATA_FOLDER/test.parquet \
    data.train_batch_size=512 \
    data.max_prompt_length=4000 \
    data.max_response_length=8000 \
    data.filter_overlong_prompts=True \
    data.truncation='error' \
    data.return_raw_chat=True \
    actor_rollout_ref.model.path=$MODEL_PATH \
    actor_rollout_ref.actor.optim.lr=1e-6 \
    actor_rollout_ref.model.use_remove_padding=True \
    actor_rollout_ref.actor.ppo_mini_batch_size=512 \
    actor_rollout_ref.actor.ppo_micro_batch_size_per_gpu=8 \
    actor_rollout_ref.actor.use_kl_loss=True \
    actor_rollout_ref.actor.kl_loss_coef=0.001 \
    actor_rollout_ref.actor.kl_loss_type=low_var_kl \
    actor_rollout_ref.actor.entropy_coeff=0 \
    actor_rollout_ref.model.use_fused_kernels=True \
    actor_rollout_ref.actor.use_dynamic_bsz=True \
    actor_rollout_ref.actor.ppo_max_token_len_per_gpu=24000 \
    actor_rollout_ref.rollout.log_prob_use_dynamic_bsz=true \
    actor_rollout_ref.rollout.log_prob_max_token_len_per_gpu=24000 \
    actor_rollout_ref.ref.log_prob_use_dynamic_bsz=true  \
    actor_rollout_ref.ref.log_prob_max_token_len_per_gpu=24000 \
    actor_rollout_ref.model.enable_gradient_checkpointing=True \
    actor_rollout_ref.actor.fsdp_config.param_offload=False \
    actor_rollout_ref.actor.fsdp_config.optimizer_offload=False \
    actor_rollout_ref.rollout.log_prob_micro_batch_size_per_gpu=32 \
    actor_rollout_ref.rollout.tensor_model_parallel_size=1 \
    actor_rollout_ref.rollout.name=sglang \
    actor_rollout_ref.rollout.gpu_memory_utilization=0.8 \
    actor_rollout_ref.rollout.n=5 \
    actor_rollout_ref.ref.log_prob_micro_batch_size_per_gpu=32 \
    actor_rollout_ref.ref.fsdp_config.param_offload=True \
    algorithm.use_kl_in_reward=False \
    trainer.critic_warmup=0 \
    trainer.logger=['console','wandb'] \
    trainer.project_name='verl_grpo_text2sql' \
    trainer.experiment_name="${data}_${tag}" \
    trainer.val_before_train=False \
    trainer.n_gpus_per_node=4 \
    trainer.nnodes=1 \
    trainer.save_freq=10 \
    trainer.test_freq=10 \
    trainer.validation_data_dir="./rollouts_sql_train/" \
    actor_rollout_ref.rollout.multi_turn.tool_config_path="$PROJECT_DIR/examples/sglang_multiturn/config/tool_config/text2sql_tool_config.yaml" \
    trainer.total_epochs=20  > train_sql.log 2>&1  $@


#> train_sql.log 2>&1