| license: apache-2.0 | |
| tags: | |
| - merge | |
| - mergekit | |
| - lazymergekit | |
| - gradientai/Llama-3-8B-Instruct-Gradient-1048k | |
| - EmbeddedLLM/Mistral-7B-Merge-14-v0.1 | |
| # RLM-7B-v1 | |
| RLM-7B-v1 is a merge of the following models using [mergekit](https://github.com/cg123/mergekit): | |
| * [gradientai/Llama-3-8B-Instruct-Gradient-1048k](https://huggingface.co/gradientai/Llama-3-8B-Instruct-Gradient-1048k) | |
| * [EmbeddedLLM/Mistral-7B-Merge-14-v0.1](https://huggingface.co/EmbeddedLLM/Mistral-7B-Merge-14-v0.1) | |
| ## 🧩 Configuration | |
| ```yaml | |
| slices: | |
| - sources: | |
| - model: gradientai/Llama-3-8B-Instruct-Gradient-1048k | |
| layer_range: [0, 32] | |
| - model: EmbeddedLLM/Mistral-7B-Merge-14-v0.1 | |
| layer_range: [0, 32] | |
| merge_method: slerp | |
| base_model: gradientai/Llama-3-8B-Instruct-Gradient-1048k | |
| parameters: | |
| t: | |
| - filter: self_attn | |
| value: [0, 0.5, 0.3, 0.7, 1] | |
| - filter: mlp | |
| value: [1, 0.5, 0.7, 0.3, 0] | |
| - value: 0.5 | |
| dtype: bfloat16 | |
| ``` |