okoge-kaz / moe-recipes
Ongoing research training Mixture of Expert models.
☆19Updated 6 months ago
Alternatives and similar repositories for moe-recipes:
Users that are interested in moe-recipes are comparing it to the libraries listed below
- ☆14Updated 6 months ago
- Mixtral-based Ja-En (En-Ja) Translation model☆19Updated 2 months ago
- Easily turn large English text datasets into Japanese text datasets using open LLMs.☆20Updated 2 months ago
- ☆59Updated 9 months ago
- Japanese LLaMa experiment☆53Updated 3 months ago
- Swallowプロジェクト 大規模言語モデル 評価スクリプト☆16Updated this week
- ☆49Updated 11 months ago
- Official implementation of "TAID: Temporally Adaptive Interpolated Distillation for Efficient Knowledge Transfer in Language Models"☆99Updated 2 months ago
- ☆42Updated last year
- LEIA: Facilitating Cross-Lingual Knowledge Transfer in Language Models with Entity-based Data Augmentation☆21Updated 11 months ago
- Example of using Epochraft to train HuggingFace transformers models with PyTorch FSDP☆12Updated last year
- Support Continual pre-training & Instruction Tuning forked from llama-recipes☆31Updated last year
- ☆83Updated last year
- Preferred Generation Benchmark☆78Updated this week
- ☆15Updated last year
- ☆48Updated 3 months ago
- Project of llm evaluation to Japanese tasks☆81Updated last month
- ☆26Updated 4 months ago
- ☆52Updated 9 months ago
- A Slack Bot for summarizing arXiv papers, powered by OpenAI LLMs.☆70Updated last year
- The robust text processing pipeline framework enabling customizable, efficient, and metric-logged text preprocessing.☆121Updated 5 months ago
- ☆22Updated last year
- Mamba training library developed by kotoba technologies☆68Updated last year
- ☆22Updated last year
- Japanese-BPEEncoder☆41Updated 3 years ago
- ☆20Updated 2 months ago
- CycleQD is a framework for parameter space model merging.☆35Updated last month
- ☆39Updated last month
- Checkpointable dataset utilities for foundation model training☆32Updated last year
- ☆125Updated last month