forked from OpenRLHF/OpenRLHF
-
Notifications
You must be signed in to change notification settings - Fork 0
/
train_ppo_llama_ray_70b.sh
executable file
·40 lines (38 loc) · 1.2 KB
/
train_ppo_llama_ray_70b.sh
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
set -x
ray job submit --address="http://127.0.0.1:8265" \
--runtime-env-json='{"working_dir": "/openrlhf"}' \
--no-wait \
-- python3 -m openrlhf.cli.train_ppo_ray \
--ref_num_nodes 1 \
--ref_num_gpus_per_node 2 \
--reward_num_nodes 1 \
--reward_num_gpus_per_node 2 \
--critic_num_nodes 1 \
--critic_num_gpus_per_node 4 \
--actor_num_nodes 1 \
--actor_num_gpus_per_node 4 \
--vllm_num_engines 2 \
--vllm_tensor_parallel_size 2 \
--pretrain meta-llama/Meta-Llama-3-70B-Instruct \
--reward_pretrain meta-llama/Meta-Llama-3-70B-Instruct \
--save_path /openrlhf/examples/checkpoint/llama-3-70b-rlhf \
--micro_train_batch_size 1 \
--train_batch_size 128 \
--micro_rollout_batch_size 2 \
--rollout_batch_size 1024 \
--max_epochs 1 \
--prompt_max_len 1024 \
--generate_max_len 1024 \
--zero_stage 3 \
--bf16 \
--actor_learning_rate 5e-7 \
--critic_learning_rate 9e-6 \
--init_kl_coef 0.01 \
--prompt_data OpenRLHF/prompt-collection-v0.1 \
--input_key context_messages \
--apply_chat_template \
--max_samples 10000 \
--normalize_reward \
--adam_offload \
--flash_attn \
--gradient_checkpointing