Commit
•
8cf827b
1
Parent(s):
03508cf
Update README.md
Browse files
README.md
CHANGED
|
@@ -14,15 +14,11 @@ In the meantime, you can check [this topic](https://huggingface.co/LWDCLS/LLM-Di
|
|
| 14 |
|
| 15 |
# Testing Environment
|
| 16 |
|
| 17 |
-
All models are loaded in Q8_0 (GGUF) using KoboldCPP 1.65 for Windows using CUDA 12. Using CuBLAS but not using mmq.
|
| 18 |
-
|
| 19 |
-
All layers are on the GPU (NVidia RTX3060 12GB).
|
| 20 |
|
| 21 |
Frontend is staging version of Silly Tavern.
|
| 22 |
|
| 23 |
-
All models are extended to 16K context length (auto rope from KCPP) with Flash Attention enabled.
|
| 24 |
-
|
| 25 |
-
Response size set to 1024 tokens max.
|
| 26 |
|
| 27 |
Fixed Seed for all tests: 123
|
| 28 |
|
|
|
|
| 14 |
|
| 15 |
# Testing Environment
|
| 16 |
|
| 17 |
+
All models are loaded in Q8_0 (GGUF) using KoboldCPP 1.65 for Windows using CUDA 12. Using CuBLAS but not using mmq. All layers are on the GPU (NVidia RTX3060 12GB).
|
|
|
|
|
|
|
| 18 |
|
| 19 |
Frontend is staging version of Silly Tavern.
|
| 20 |
|
| 21 |
+
All models are extended to 16K context length (auto rope from KCPP) with Flash Attention enabled. Response size set to 1024 tokens max.
|
|
|
|
|
|
|
| 22 |
|
| 23 |
Fixed Seed for all tests: 123
|
| 24 |
|