NousResearch/Nous-Capybara-34B, migtissera/Tess-M-v1.2 and migtissera/Tess-M-v1.3 merged with a new, experimental implementation of "dare ties" via mergekit. See:
Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch
https://github.com/yule-BUAA/MergeLM
https://github.com/cg123/mergekit/tree/dare-tokenizer
Highly experimental and still being tested! But this should yield a better merge than a typical linear/slerp merge or even a ties merge.
Merged with the following config, and the tokenizer from Yi Llamafied:
models:
- model: /home/alpha/Storage/Models/Raw/larryvrh_Yi-34B-200K-Llamafied
# no parameters necessary for base model
- model: /home/alpha/Storage/Models/Raw/migtissera_Tess-M-v1.3
parameters:
weight: 0.50
density: 0.56
- model: /home/alpha/Storage/Models/Raw/migtissera_Tess-M-v1.2
parameters:
weight: 0.20
density: 0.50
- model: /home/alpha/Storage/Models/Raw/Nous-Capybara-34B
parameters:
weight: 0.50
density: 0.56
merge_method: dare_ties
base_model: /home/alpha/Storage/Models/Raw/larryvrh_Yi-34B-200K-Llamafied
parameters:
int8_mask: true
dtype: bfloat16
Tess 1.2 (at a low weight) and 1.3 were used because, according to the trainer, they were trained on different datasets: https://migel.substack.com/p/learnings-from-training-tess
I chose not to include other finetunes, such as Dolphin, because they aren't trained on the 200K base.
Prompt template: Orca-Vicuna
SYSTEM: {system_message}
USER: {prompt}
ASSISTANT:
Being a Yi model, try disabling the BOS token and/or running a lower temperature with MinP if output doesn't seem right.
Sometimes the model "spells out" the stop token as </s> like Capybara, so you may need to add </s> as an additional stopping condition.
Credits:
https://github.com/cg123/mergekit/tree/dare-tokenizer
https://huggingface.co./NousResearch/Nous-Capybara-34B/
https://huggingface.co./migtissera/Tess-M-v1.2
https://huggingface.co./migtissera/Tess-M-v1.3
- Downloads last month
- 6