cooper12121 / llama3-8x8b-MoELinks
Copy the MLP of llama3 8 times as 8 experts , created a router with random initialization,add load balancing loss to construct an 8x8b MoE model based on llama3.
☆27Updated last year
Alternatives and similar repositories for llama3-8x8b-MoE
Users that are interested in llama3-8x8b-MoE are comparing it to the libraries listed below
Sorting:
- ☆49Updated last year
- ☆49Updated last year
- [ICML'24] The official implementation of “Rethinking Optimization and Architecture for Tiny Language Models”☆123Updated 8 months ago
- [ICML 2025] |TokenSwift: Lossless Acceleration of Ultra Long Sequence Generation☆113Updated 4 months ago
- ☆89Updated 4 months ago
- Automatic prompt optimization framework for multi-step agent tasks.☆34Updated 11 months ago
- SELF-GUIDE: Better Task-Specific Instruction Following via Self-Synthetic Finetuning. COLM 2024 Accepted Paper☆33Updated last year
- A highly capable 2.4B lightweight LLM using only 1T pre-training data with all details.☆217Updated 2 months ago
- ☆36Updated last year
- ☆95Updated 10 months ago
- FuseAI Project☆87Updated 8 months ago
- ☆49Updated 2 months ago
- Implementation of the LongRoPE: Extending LLM Context Window Beyond 2 Million Tokens Paper☆150Updated last year
- Reformatted Alignment☆112Updated last year
- Parameter-Efficient Sparsity Crafting From Dense to Mixture-of-Experts for Instruction Tuning on General Tasks (EMNLP'24)☆147Updated last year
- [ICLR 2025] LongPO: Long Context Self-Evolution of Large Language Models through Short-to-Long Preference Optimization☆40Updated 7 months ago
- Skywork-MoE: A Deep Dive into Training Techniques for Mixture-of-Experts Language Models☆137Updated last year
- This is a personal reimplementation of Google's Infini-transformer, utilizing a small 2b model. The project includes both model and train…☆58Updated last year
- Scaling Preference Data Curation via Human-AI Synergy☆113Updated 3 months ago
- ☆115Updated last year
- [ACL 2024] LLM2LLM: Boosting LLMs with Novel Iterative Data Enhancement☆190Updated last year
- ☆104Updated 10 months ago
- ☆90Updated 11 months ago
- Delta-CoMe can achieve near loss-less 1-bit compressin which has been accepted by NeurIPS 2024☆57Updated 10 months ago
- Code for "Critique Fine-Tuning: Learning to Critique is More Effective than Learning to Imitate" [COLM 2025]☆173Updated 3 months ago
- The source code and dataset mentioned in the paper Seal-Tools: Self-Instruct Tool Learning Dataset for Agent Tuning and Detailed Benchmar…☆51Updated 11 months ago
- code for Scaling Laws of RoPE-based Extrapolation☆73Updated last year
- ☆100Updated 4 months ago
- ☆65Updated 10 months ago
- Official completion of “Training on the Benchmark Is Not All You Need”.☆36Updated 9 months ago