llama-3-moe-3x8b / mergekit_moe_config.yml
yrju's picture
Upload folder using huggingface_hub
03547a8 verified
base_model: meta-llama/Meta-Llama-3-8B-Instruct
gate_mode: hidden # one of "hidden", "cheap_embed", or "random"
dtype: bfloat16 # output dtype (float32, float16, or bfloat16)
## (optional)
# experts_per_token: 2
experts:
- source_model: meta-llama/Meta-Llama-3-8B-Instruct
positive_prompts:
- "chat"
- "assistant"
- "tell me"
- "explain"
- "I want"
## (optional)
# negative_prompts:
# - "This is a prompt expert_model_1 should not be used for"
- source_model: codellama/CodeLlama-7b-Instruct-hf
positive_prompts:
- "code"
- "python"
- "javascript"
- "programming"
- "algorithm"
- "C#"
- "C++"
- "debug"
- "runtime"
- "html"
- "command"
- "nodejs"
- source_model: meta-math/MetaMath-Mistral-7B
positive_prompts:
- "reason"
- "math"
- "mathematics"
- "solve"
- "count"
- "calculate"
- "arithmetic"
- "algebra"