Hugging Face's logo

Kquant03
/
FrankenDPO-4x7B-GGUF

GGUF
English
Merge
Inference Endpoints
Model card Files Community
1
Kquant03 commited on
Commit
cad5879
·
verified ·
1 Parent(s): 0aa2f5b

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +4 -1
README.md CHANGED
@@ -6,13 +6,16 @@ tags:
6
  - merge
7
  ---
8
  ![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/6589d7e6586088fd2784a12c/7JsqBt8QRiZmcMh-ameqH.jpeg)
9
- # It's alive!!!! Half the size and better on GSM8k and Winogrande than Mixtral Instruct 8x 7B!
10
 
11
  A frankenMoE using only DPO models. To be used with Chat-instruct mode enabled.
12
 
13
  ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6589d7e6586088fd2784a12c/wGRcusncUd-mCdksvYckY.png)
14
 
15
  ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6589d7e6586088fd2784a12c/rx1GfLMEIP3T-r3bxqW9r.png)
 
 
 
16
  ## Provided files
17
 
18
  | Name | Quant method | Bits | Size | Max RAM required | Use case |
 
6
  - merge
7
  ---
8
  ![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/6589d7e6586088fd2784a12c/7JsqBt8QRiZmcMh-ameqH.jpeg)
9
+ # It's alive!!!! Half the size and better on GSM8k and Winogrande than Mixtral Instruct 8x 7B! Also rank 6 on Ayumi's ERP Bench!
10
 
11
  A frankenMoE using only DPO models. To be used with Chat-instruct mode enabled.
12
 
13
  ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6589d7e6586088fd2784a12c/wGRcusncUd-mCdksvYckY.png)
14
 
15
  ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6589d7e6586088fd2784a12c/rx1GfLMEIP3T-r3bxqW9r.png)
16
+
17
+ ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6589d7e6586088fd2784a12c/3F_Sm5He9AlsfyRcvcZqk.png)
18
+
19
  ## Provided files
20
 
21
  | Name | Quant method | Bits | Size | Max RAM required | Use case |