Commit
•
2e63d25
1
Parent(s):
34bf352
Update README.md
Browse files
README.md
CHANGED
|
@@ -3,6 +3,11 @@ license: other
|
|
| 3 |
datasets:
|
| 4 |
- jondurbin/airoboros-gpt4-1.4.1
|
| 5 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
|
| 7 |
### Overview
|
| 8 |
|
|
@@ -42,4 +47,4 @@ I am purposingly leaving this license ambiguous (other than the fact you must co
|
|
| 42 |
|
| 43 |
Your best bet is probably to avoid using this commercially due to the OpenAI API usage.
|
| 44 |
|
| 45 |
-
Either way, by using this model, you agree to completely indemnify me.
|
|
|
|
| 3 |
datasets:
|
| 4 |
- jondurbin/airoboros-gpt4-1.4.1
|
| 5 |
---
|
| 6 |
+
### 2.4-bit Exllama v2 quant of airoboros-gpt4-1.4.1 using new quant method
|
| 7 |
+
|
| 8 |
+
Simple quantization of original model. This model should fit on a single 24 GB VRAM GPU where Exlalama v2 is supported.
|
| 9 |
+
Should also support full 4096 context on a single GPU, without dsktop apps also running on the same GPU. Ideally,
|
| 10 |
+
the GPU would be completely empty of any desktop or apps.
|
| 11 |
|
| 12 |
### Overview
|
| 13 |
|
|
|
|
| 47 |
|
| 48 |
Your best bet is probably to avoid using this commercially due to the OpenAI API usage.
|
| 49 |
|
| 50 |
+
Either way, by using this model, you agree to completely indemnify me.
|