Llemma 7B muInstruct Camel Math
This model fine-tunes Llemma 7B on EleutherAI muInstruct and camel-ai/math. It changes the mathematical base into an instruction-following research model.
Research release
The prompt and end token matter
The documented prompt contains Input: followed by the question, then a blank line and Response:. A training error changed the end-of-sequence token ID to 0 instead of the usual Llama 2 value of 2. Load the released configuration rather than imposing the stop token from the base model.llemma_7b_muinstruct_camelmath model card
A trained adaptation with inherited terms
The card declares Apache 2.0 for the adaptation. Its base descends from Code Llama under Llama 2 terms. Check the upstream license before distribution.llemma_7b_muinstruct_camelmath model cardllemma_7b model card
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Research release |
| Access | Downloadable |
| Model identifiers | EleutherAI/llemma_7b_muinstruct_camelmath |
| Context window | 4,096 tokens Maximum position count in the released model configuration. llemma_7b_muinstruct_camelmath model configuration |
| Nominal parameters | 7Bllemma_7b_muinstruct_camelmath model card |
| Training datasets | muInstruct and camel-ai/mathllemma_7b_muinstruct_camelmath model card |
| End-of-sequence token ID | 0llemma_7b_muinstruct_camelmath model configuration |
| License notes | Card: Apache 2.0; underlying Code Llama: Llama 2llemma_7b_muinstruct_camelmath model cardllemma_7b model card |
Related models
- Llemma 7B
EleutherAI ยท Llemma
A 7B mathematical language model adapted from Meta Code Llama.
Launch date Research releaseDownloadable