Downloads · 30 days
0
wzx111/Qwen3-1.7B-GRPO-math
Qwen3-1.7B-GRPO-math is a machine learning model from wzx111. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
本模型由 VERL (Ray PPO/ADPO) 训练生成,并在训练结束自动上传。
Downloads · 30 days
0
Access
Public
Updated Nov 29, 2025
Repo size
20.7 GB
Likes
0
Public
Click a slice to open those files.
.pt10.3 GB · 100%
From the Hugging Face model README
本模型由 VERL (Ray PPO/ADPO) 训练生成,并在训练结束自动上传。
Qwen/Qwen3-1.7Bqwen3-1.7b-grpo-wzx-4gpu2bfloat16grpo (use_q_centering=True, use_kde_advantage=False, scale_rewards=False)8, max_num_seqs=300, gpu_memory_utilization=0.424, ppo_micro_batch_size_per_gpu=6Nonefrom transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("wzx111/Qwen3-1.7B-GRPO-math")
tokenizer = AutoTokenizer.from_pretrained("wzx111/Qwen3-1.7B-GRPO-math")
print(model.config)
global_step_175/actor/请遵循上游基础模型与数据集的许可条款。本模型卡由 VERL 自动生成。