Commit
•
0547f33
1
Parent(s):
c6b28ef
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,14 @@
|
|
| 1 |
---
|
| 2 |
license: other
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: other
|
| 3 |
---
|
| 4 |
+
This model is a merge between the finetuned model named airoboros-gpt4-1.4 and a Lora named GamerUntouchStorytelling
|
| 5 |
+
https://huggingface.co/jondurbin/airoboros-13b-gpt4-1.4
|
| 6 |
+
https://huggingface.co/GamerUntouch/Storytelling-LLaMa-LoRAs
|
| 7 |
+
|
| 8 |
+
Airoboros is a great base model as it understands the requets well and has a great english prose.
|
| 9 |
+
The problem was that for Storytelling purposes, it was really blank and the model never added something new to the table.
|
| 10 |
+
Mixing this finetuned model with a Storytelling Lora really made it more creative and interesting, I hope you'll enjoy it :)
|
| 11 |
+
|
| 12 |
+
This gptq safetensor has all the implementations in it (true_sequential + act_order + groupsize 32)
|
| 13 |
+
|
| 14 |
+
Run it with exllama or exllama_hf, this marvelous feature can load everything!
|