Update README.md
Browse files
README.md
CHANGED
|
@@ -68,7 +68,10 @@ This model has a max context window size of 128k tokens. By default, a
|
|
| 68 |
context window size of 4096 tokens is used. You can use a larger context
|
| 69 |
window by passing the `-c 8192` flag. The software currently has
|
| 70 |
limitations in its llama v3.1 support that may prevent scaling to the
|
| 71 |
-
full 128k size.
|
|
|
|
|
|
|
|
|
|
| 72 |
|
| 73 |
On GPUs with sufficient RAM, the `-ngl 999` flag may be passed to use
|
| 74 |
the system's NVIDIA or AMD GPU(s). On Windows, only the graphics card
|
|
|
|
| 68 |
context window size of 4096 tokens is used. You can use a larger context
|
| 69 |
window by passing the `-c 8192` flag. The software currently has
|
| 70 |
limitations in its llama v3.1 support that may prevent scaling to the
|
| 71 |
+
full 128k size. See our
|
| 72 |
+
[Phi-3-medium-128k-instruct-llamafile](https://huggingface.co/Mozilla/Phi-3-medium-128k-instruct-llamafile)
|
| 73 |
+
repository for llamafiles that are known to work with a 128kb context
|
| 74 |
+
size.
|
| 75 |
|
| 76 |
On GPUs with sufficient RAM, the `-ngl 999` flag may be passed to use
|
| 77 |
the system's NVIDIA or AMD GPU(s). On Windows, only the graphics card
|