S1quence

icon indicating a website link https://iic.sdu.edu.cn/

icon company Shandong University icon location Jimo district, Qingdao, Shandong icon location Want to buy PS5, Xbox Game Pass Ultimate, and Switch.

Results 3 issues of S1quence

Some Bugs in JetMoE

4 comment

### System Info transformers version: 4.43.0.dev0 (installed from source) ### Who can help? @ArthurZucker ### Information - [X] The official example scripts - [ ] My own modified scripts ###...

Fixed bugs in JetMoE models.

4 comment

# What does this PR do? Fixes issue #31791. 1. Fixed bugs in JetMoE casual language model and sequence classification model. 2. The original code will report errors when we...

Embedding checkpoint size mismatch when using peft on DeepSeek-R1-Distill-Qwen-1.5B.

## Summary I encountered a size mismatch problem like in https://github.com/EleutherAI/lm-evaluation-harness/issues/1707#issue-2244839671 when evaluating peft model (DeepSeek-R1-Distill-Qwen-1.5B), but the root cause is due to the inherent mismatch with model's `vocab_size` and...