merge
This is a merge of pre-trained language models created using mergekit.
Merge Details
- This model is an attempt at making a smart rp model with the finesse of Epiculous/Fett-uccine-7B.
- From limited testing i've found it to be my favourite of my personal 7B models. It stays pretty coherent at 8k+ ctx.
- I like to use "Alpaca" format with "Universal-Light" for longer messages. Switching to ChatML causes the messages to be much shorter? I haven't a clue why but sometimes it's nice.
- It doesn't seem to show many issues but i'd be willing to try to fix any problems or bugs as it shows some potential.
Merge Method
This model was merged using the DARE TIES merge method using OpenPipe/mistral-ft-optimized-1227 as a base.
Models Merged
The following models were included in the merge:
- Epiculous/Fett-uccine-7B
- eren23/ogno-monarch-jaskier-merge-7b-OH-PREF-DPO-v2
- ChaoticNeutrals/Eris_7B
Configuration
The following YAML configuration was used to produce this model:
models:
- model: OpenPipe/mistral-ft-optimized-1227
# No parameters necessary for base model
- model: Epiculous/Fett-uccine-7B
parameters:
density: 0.53
weight: 0.4
- model: eren23/ogno-monarch-jaskier-merge-7b-OH-PREF-DPO-v2
parameters:
density: 0.53
weight: 0.35
- model: ChaoticNeutrals/Eris_7B
parameters:
density: 0.53
weight: 0.25
merge_method: dare_ties
base_model: OpenPipe/mistral-ft-optimized-1227
parameters:
int8_mask: true
dtype: bfloat16
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
Metric | Value |
---|---|
Avg. | 71.66 |
AI2 Reasoning Challenge (25-Shot) | 68.77 |
HellaSwag (10-Shot) | 87.33 |
MMLU (5-Shot) | 63.65 |
TruthfulQA (0-shot) | 71.91 |
Winogrande (5-shot) | 80.82 |
GSM8k (5-shot) | 57.47 |
- Downloads last month
- 76
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.
Model tree for saishf/Fett-Eris-Mix-7B
Merge model
this model
Evaluation results
- normalized accuracy on AI2 Reasoning Challenge (25-Shot)test set Open LLM Leaderboard68.770
- normalized accuracy on HellaSwag (10-Shot)validation set Open LLM Leaderboard87.330
- accuracy on MMLU (5-Shot)test set Open LLM Leaderboard63.650
- mc2 on TruthfulQA (0-shot)validation set Open LLM Leaderboard71.910
- accuracy on Winogrande (5-shot)validation set Open LLM Leaderboard80.820
- accuracy on GSM8k (5-shot)test set Open LLM Leaderboard57.470