AIM Merged Checkpoints (Baseline W/O AIM)
Collection
The full set of checkpoints merged without AIM, used in Activation Informed Merging (AIM) merging paper experiments. • 22 items • Updated
AIM Paper Checkpoints Uploaded For Replication
This repository includes one of the checkpoints used in the paper "Activation-Informed Merging of Large Language Models". Specifics of this model are as follows:
- Merging Method: dare_linear
- Models Used In Merging
- Base Model: unsloth/llama-2-13b
- Code: layoric/llama-2-13b-code-alpaca
- Math: vanillaOVO/WizardMath-13B-V1.0
- Instruction Tuned: WizardLMTeam/WizardLM-13B-V1.2
- AIM: False
Benchmark results and paper details can be found at the official GitHub.
- Downloads last month
- 14
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for ahn1376/DARETaskArithmetic___Code-Math-Instruction_Tuned
Merge model
this model