Update README.md
Browse files
README.md
CHANGED
|
@@ -1,4 +1,5 @@
|
|
| 1 |
---
|
|
|
|
| 2 |
library_name: transformers
|
| 3 |
tags:
|
| 4 |
- 4-bit
|
|
@@ -10,6 +11,32 @@ pipeline_tag: text-generation
|
|
| 10 |
inference: false
|
| 11 |
quantized_by: Suparious
|
| 12 |
---
|
| 13 |
-
#
|
| 14 |
|
| 15 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
library_name: transformers
|
| 4 |
tags:
|
| 5 |
- 4-bit
|
|
|
|
| 11 |
inference: false
|
| 12 |
quantized_by: Suparious
|
| 13 |
---
|
| 14 |
+
# dawn17/MistarlingMaid-2x7B-base AWQ
|
| 15 |
|
| 16 |
+
- Model creator: [dawn17](https://huggingface.co/dawn17)
|
| 17 |
+
- Original model: [MistarlingMaid-2x7B-base](https://huggingface.co/dawn17/MistarlingMaid-2x7B-base)
|
| 18 |
+
|
| 19 |
+
## Model Summary
|
| 20 |
+
|
| 21 |
+
base_model: /Users/dawn/git/models/Mistral-7B-Instruct-v0.2
|
| 22 |
+
gate_mode: hidden # one of "hidden", "cheap_embed", or "random"
|
| 23 |
+
dtype: bfloat16 # output dtype (float32, float16, or bfloat16)
|
| 24 |
+
experts:
|
| 25 |
+
- source_model: /Users/dawn/git/models/Silicon-Maid-7B
|
| 26 |
+
positive_prompts:
|
| 27 |
+
- "roleplay"
|
| 28 |
+
- source_model: /Users/dawn/git/models/Starling-LM-7B-beta
|
| 29 |
+
positive_prompts:
|
| 30 |
+
- "chat"
|
| 31 |
+
|
| 32 |
+
- [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
|
| 33 |
+
|
| 34 |
+
| Metric |Value|
|
| 35 |
+
|---------------------------------|----:|
|
| 36 |
+
|Avg. |68.01|
|
| 37 |
+
|AI2 Reasoning Challenge (25-Shot)|67.49|
|
| 38 |
+
|HellaSwag (10-Shot) |84.76|
|
| 39 |
+
|MMLU (5-Shot) |62.62|
|
| 40 |
+
|TruthfulQA (0-shot) |58.93|
|
| 41 |
+
|Winogrande (5-shot) |78.22|
|
| 42 |
+
|GSM8k (5-shot) |56.03|
|