Commit
•
94d21b4
1
Parent(s):
148c9d6
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,8 +1,6 @@
|
|
| 1 |
---
|
| 2 |
license: cc-by-4.0
|
| 3 |
---
|
| 4 |
-
Repository Description:
|
| 5 |
-
|
| 6 |
Introducing PirateTalk-13b-v1-GGUF-Q8, a CPU-optimized iteration of the original PirateTalk-13b-v1 model, founded on the robust 13b Llama 2 Chat architecture. This version has been meticulously quantized to 8 bits utilizing the llama.cpp utility, making it apt for CPU inferences without the necessity of a GPU. With a minimal requirement of 16 GB RAM, the performance of this model is contingent on the CPU speed, which may range from slow to reasonable. Despite the transition from fp16 to 8-bit quantization, a discernible piratical flair is retained, albeit with a modest compromise in vernacular quality.
|
| 7 |
|
| 8 |
Objective:
|
|
|
|
| 1 |
---
|
| 2 |
license: cc-by-4.0
|
| 3 |
---
|
|
|
|
|
|
|
| 4 |
Introducing PirateTalk-13b-v1-GGUF-Q8, a CPU-optimized iteration of the original PirateTalk-13b-v1 model, founded on the robust 13b Llama 2 Chat architecture. This version has been meticulously quantized to 8 bits utilizing the llama.cpp utility, making it apt for CPU inferences without the necessity of a GPU. With a minimal requirement of 16 GB RAM, the performance of this model is contingent on the CPU speed, which may range from slow to reasonable. Despite the transition from fp16 to 8-bit quantization, a discernible piratical flair is retained, albeit with a modest compromise in vernacular quality.
|
| 5 |
|
| 6 |
Objective:
|