Update README.md
Browse files
README.md
CHANGED
|
@@ -1,58 +0,0 @@
|
|
| 1 |
-
---
|
| 2 |
-
base_model:
|
| 3 |
-
- arcee-train/arcee-maz-mistral-24b-v3
|
| 4 |
-
library_name: transformers
|
| 5 |
-
tags:
|
| 6 |
-
- mergekit
|
| 7 |
-
- merge
|
| 8 |
-
|
| 9 |
-
---
|
| 10 |
-
# Mistral-24B-MazMerge-V3
|
| 11 |
-
|
| 12 |
-
This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
|
| 13 |
-
|
| 14 |
-
## Merge Details
|
| 15 |
-
### Merge Method
|
| 16 |
-
|
| 17 |
-
This model was merged using the [Arcee Fusion](https://arcee.ai) merge method using /workspace/models/Mistral-24B-MazMerge-V1 as a base.
|
| 18 |
-
|
| 19 |
-
### Models Merged
|
| 20 |
-
|
| 21 |
-
The following models were included in the merge:
|
| 22 |
-
* [arcee-train/arcee-maz-mistral-24b-v3](https://huggingface.co/arcee-train/arcee-maz-mistral-24b-v3)
|
| 23 |
-
|
| 24 |
-
### Configuration
|
| 25 |
-
|
| 26 |
-
The following YAML configuration was used to produce this model:
|
| 27 |
-
|
| 28 |
-
```yaml
|
| 29 |
-
base_model: /workspace/models/Mistral-24B-MazMerge-V1
|
| 30 |
-
dtype: float32
|
| 31 |
-
merge_method: arcee_fusion
|
| 32 |
-
out_dtype: bfloat16
|
| 33 |
-
slices:
|
| 34 |
-
- sources:
|
| 35 |
-
- layer_range: [0, 40]
|
| 36 |
-
model: arcee-train/arcee-maz-mistral-24b-v3
|
| 37 |
-
- layer_range: [0, 40]
|
| 38 |
-
model: /workspace/models/Mistral-24B-MazMerge-V1
|
| 39 |
-
tokenizer:
|
| 40 |
-
source: arcee-train/arcee-maz-mistral-24b-v3
|
| 41 |
-
```
|
| 42 |
-
|
| 43 |
-
## Benchmark Results
|
| 44 |
-
|
| 45 |
-
| Groups | Version | Filter | n-shot | Metric | | Value | | Stderr |
|
| 46 |
-
| ----------------------- | ------- | ------ | ------ | ----------------------- | --- | -----: | --- | ------ |
|
| 47 |
-
| leaderboard | 1 | none | | acc | ↑ | 0.6025 | ± | 0.0045 |
|
| 48 |
-
| | | none | | acc_norm | ↑ | 0.5918 | ± | 0.0051 |
|
| 49 |
-
| - leaderboard_ifeval | | none | | inst_level_loose_acc | ↑ | 0.8525 | ± | N/A |
|
| 50 |
-
| | | none | | inst_level_strict_acc | ↑ | 0.8369 | ± | N/A |
|
| 51 |
-
| | | none | | prompt_level_loose_acc | ↑ | 0.7930 | ± | 0.0174 |
|
| 52 |
-
| | | none | | prompt_level_strict_acc | ↑ | 0.7745 | ± | 0.0180 |
|
| 53 |
-
| - leaderboard_bbh | N/A | none | | acc_norm | ↑ | 0.6504 | ± | 0.0058 |
|
| 54 |
-
| - leaderboard_gpqa | N/A | none | | acc_norm | ↑ | 0.3666 | ± | 0.0140 |
|
| 55 |
-
| - leaderboard_math_hard | N/A | none | | exact_match | ↑ | 0.3860 | ± | 0.0123 |
|
| 56 |
-
| - leaderboard_musr | N/A | none | | acc_norm | ↑ | 0.5000 | ± | 0.0180 |
|
| 57 |
-
| - leaderboard_mmlu_pro | 0.1 | none | 5 | acc | ↑ | 0.6025 | ± | 0.0045 |
|
| 58 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|