hyunwoongko / nanorlhfLinks
nanorlhf: from-scratch journey into how LLMs and RLHF really work.
☆27Updated 3 weeks ago
Alternatives and similar repositories for nanorlhf
Users that are interested in nanorlhf are comparing it to the libraries listed below
Sorting:
- ☆64Updated 3 months ago
- LLM 모델의 외국어 토큰 생성을 막는 코드 구현☆80Updated 2 months ago
- Kor-IR: Korean Information Retrieval Benchmark☆88Updated last year
- ☆36Updated 2 years ago
- StrategyQA 데이터 세트 번역☆22Updated last year
- ☆12Updated 10 months ago
- 구글에서 발표한 Chain-of-Thought Reasoning without Prompting을 코드로 구현한 레포입니다.☆67Updated last year
- Korean Sentence Embedding Model Performance Benchmark for RAG☆48Updated 9 months ago
- ☆108Updated 3 months ago
- AutoRAG example about benchmarking Korean embeddings.☆41Updated last year
- huggingface에 있는 한국어 데이터 세트☆30Updated last year
- ☆48Updated last year
- bpe based korean t5 model for text-to-text unified framework☆63Updated last year
- 한국어 심리 상담 데이터셋☆78Updated 2 years ago
- Official repository for KoMT-Bench built by LG AI Research☆69Updated last year
- 자체 구축한 한국어 평가 데이터셋을 이용한 한국어 모델 평가☆31Updated last year
- #인권코퍼스☆32Updated 2 years ago
- Performs benchmarking on two Korean datasets with minimal time and effort.☆43Updated 2 weeks ago
- Reward Model을 이용하여 언어모델의 답변을 평가하기☆29Updated last year
- A loader that lets you try running LLMs built for WebGPU.☆29Updated last year
- hwpxlib 패키지 python에서 쉽게 사용 할수 있게 만든 github repo 입니다.☆35Updated 7 months ago
- Forked repo from https://github.com/EleutherAI/lm-evaluation-harness/commit/1f66adc☆80Updated last year
- Liner LLM Meetup archive☆71Updated last year
- 한국어 언어모델 다분야 사고력 벤치마크☆198Updated last year
- hwplib 패키지 python에서 쉽게 사용 할수 있게 만든 github repo 입니다.☆50Updated 7 months ago
- ☆38Updated 7 months ago
- This repository aims to develop CoT Steering based on CoT without Prompting. It focuses on enhancing the model’s latent reasoning capabil…☆113Updated 4 months ago
- Ko-Arena-Hard-Auto: An automatic LLM benchmark for Korean☆23Updated 6 months ago
- Make running benchmark simple yet maintainable, again. Now only supports Korean-based cross-encoder.☆21Updated 3 weeks ago
- ☆21Updated 3 years ago