Skip to content

GitLab

Switch branch/tag
Find file Normal viewHistoryPermalink
run_deepseek_megatron_ckpt.sh 3.97 KB
Newer Older
jerrrrry's avatar
Initial commit  
jerrrrry committed
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
86
87
88
89
90
91 
set -x

# the config file used: verl/trainer/main_ppo/config/ppo_megatron_trainer.yaml

huggingface-cli download deepseek-ai/deepseek-llm-7b-chat

export VLLM_ATTENTION_BACKEND=XFORMERS

python3 -m verl.trainer.main_ppo --config-path=config \
    --config-name='ppo_megatron_trainer.yaml'\
    algorithm.adv_estimator=gae \
    data.train_files=$HOME/data/gsm8k/train.parquet \
    data.val_files=$HOME/data/gsm8k/test.parquet \
    data.train_batch_size=1024 \
    data.max_prompt_length=512 \
    data.max_response_length=512 \
    actor_rollout_ref.model.path=deepseek-ai/deepseek-llm-7b-chat \
    actor_rollout_ref.actor.optim.lr=2e-6 \
    actor_rollout_ref.actor.ppo_mini_batch_size=256 \
    actor_rollout_ref.actor.ppo_micro_batch_size_per_gpu=4 \
    actor_rollout_ref.actor.megatron.pipeline_model_parallel_size=2 \
    actor_rollout_ref.actor.megatron.tensor_model_parallel_size=4 \
    actor_rollout_ref.actor.use_kl_loss=False \
    actor_rollout_ref.rollout.log_prob_micro_batch_size_per_gpu=8 \
    actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
    actor_rollout_ref.rollout.name=vllm \
    actor_rollout_ref.rollout.gpu_memory_utilization=0.5 \
    actor_rollout_ref.ref.log_prob_micro_batch_size_per_gpu=16 \
    actor_rollout_ref.ref.megatron.pipeline_model_parallel_size=2 \
    actor_rollout_ref.ref.megatron.tensor_model_parallel_size=4 \
    critic.optim.lr=2e-5 \
    critic.model.path=deepseek-ai/deepseek-llm-7b-chat \
    critic.model.enable_gradient_checkpointing=False \
    critic.ppo_micro_batch_size_per_gpu=4 \
    critic.megatron.pipeline_model_parallel_size=2 \
    critic.megatron.tensor_model_parallel_size=4 \
    algorithm.use_kl_in_reward=True \
    algorithm.kl_ctrl.kl_coef=0.001 \
    trainer.critic_warmup=0 \
    trainer.logger=['console','wandb'] \
    trainer.project_name='verl_megatron_checkpoint' \
    trainer.experiment_name='deepseek_megatron_checkpoint_saveload' \
    trainer.n_gpus_per_node=16 \
    trainer.nnodes=1 \
    trainer.save_freq=50 \
    trainer.test_freq=1 \
    trainer.total_epochs=15 \
    trainer.total_training_steps=50 $@


python3 -m verl.trainer.main_ppo --config-path=config \
    --config-name='ppo_megatron_trainer.yaml'\
    algorithm.adv_estimator=gae \
    data.train_files=$HOME/data/gsm8k/train.parquet \
    data.val_files=$HOME/data/gsm8k/test.parquet \
    data.train_batch_size=1024 \
    data.max_prompt_length=512 \
    data.max_response_length=512 \
    actor_rollout_ref.model.path=deepseek-ai/deepseek-llm-7b-chat \
    actor_rollout_ref.actor.optim.lr=2e-6 \
    actor_rollout_ref.actor.ppo_mini_batch_size=256 \
    actor_rollout_ref.actor.ppo_micro_batch_size_per_gpu=4 \
    actor_rollout_ref.actor.megatron.pipeline_model_parallel_size=2 \
    actor_rollout_ref.actor.megatron.tensor_model_parallel_size=4 \
    actor_rollout_ref.actor.use_kl_loss=False \
    actor_rollout_ref.rollout.log_prob_micro_batch_size_per_gpu=8 \
    actor_rollout_ref.rollout.tensor_model_parallel_size=2 \
    actor_rollout_ref.rollout.name=vllm \
    actor_rollout_ref.rollout.gpu_memory_utilization=0.5 \
    actor_rollout_ref.ref.log_prob_micro_batch_size_per_gpu=16 \
    actor_rollout_ref.ref.megatron.pipeline_model_parallel_size=2 \
    actor_rollout_ref.ref.megatron.tensor_model_parallel_size=4 \
    critic.optim.lr=2e-5 \
    critic.model.path=deepseek-ai/deepseek-llm-7b-chat \
    critic.model.enable_gradient_checkpointing=False \
    critic.ppo_micro_batch_size_per_gpu=4 \
    critic.megatron.pipeline_model_parallel_size=2 \
    critic.megatron.tensor_model_parallel_size=4 \
    algorithm.use_kl_in_reward=True \
    algorithm.kl_ctrl.kl_coef=0.001 \
    trainer.critic_warmup=0 \
    trainer.logger=['console','wandb'] \
    trainer.project_name='verl_megatron_checkpoint' \
    trainer.experiment_name='deepseek_megatron_checkpoint_saveload' \
    trainer.n_gpus_per_node=16 \
    trainer.nnodes=1 \
    trainer.resume_mode=auto \
    trainer.save_freq=-1 \
    trainer.test_freq=1 \
    trainer.total_epochs=15 \
    trainer.total_training_steps=150 $@