MiniOneRec

MiniOneRec copied to clipboard

Hi, thank you for the great work and contribution to the open-source community! I would like to know more details about the models in the transfer capability experiment: **Qwen-Text**: The...

cookieminions

Reproduction Results of SFT and RL Are Lower Than Reported — Any Suggestions?

6

Firstly, thank you for the great work! We have reproduced the experiments using both SFT and RL, following the exact hyperparameters provided in the paper. The experiments were conducted on...

maobenz

lovemaoli

大模型权重

2

作者您好，可以放一下您使用的模型权重吗，这里库太多了，版本间差异太大，谢谢！ 同时是否有小一点且适合的大模型权重呢？ def load_qwen_model(model_path): print("Loading Qwen Model:", model_path) tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True) model = AutoModel.from_pretrained( model_path, trust_remote_code=True, torch_dtype=torch.float16, low_cpu_mem_usage=True ) return tokenizer, model

1819707238

运行bash rl.sh报错

3

运行bash rl.sh \ --model_path Qwen/Qwen3-0.6B \ --output_dir output_dir 报错Exception ignored in: Traceback (most recent call last): File "/root/.local/share/mamba/envs/onerec3/lib/python3.10/site-packages/deepspeed/runtime/engine.py", line 528, in __del__ self.destroy() File "/root/.local/share/mamba/envs/onerec3/lib/python3.10/site-packages/deepspeed/runtime/engine.py", line 532, in destroy self.optimizer.destroy()...

gzj-ai

Hi, thanks for the solid work. I'm curious why padding is applied from the left side in the code, such as `tokenizer.padding_side = "left"`. Is there any advantage of left...

Run542968

运行bash generate_indices_plus.sh 报错Loading dataset from ../data/Amazon18/Industrial_and_Scientific/Industrial_and_Scientific.emb-qwen-td.npy Loaded embeddings shape: (26, 2048) Embeddings stats - min: -55.880207, max: 59.651443, mean: -0.005308 Building model with e_dim=2048 (Must match input dim 2048)... Loading...

gzj-ai

Qwen2.5-0.5B outputs JSON-like text instead of SID tokens during evaluation

3

Hi authors, thank you for releasing MiniOneRec! I am trying to reproduce the SFT results, but using a smaller backbone: Base model: Qwen2.5-0.5B-Instruct Dataset: Amazon-18 Musical Instruments (processed as described)...

hupeiyu21

Customized datasets

1

I want to train on a customized dataset. What inputs do I need? And how can I modify the code to quickly adapt my customized dataset? Thanks

Eddie-cc

How to control the target category for generative recommendations?

3

Thanks for your great work! I'd like to know if there's a way to control which item categories are generated during the recommendation process. For example, if I want to...

diodes-zhang