Commit
•
007bedb
1
Parent(s):
3d93d82
Update README.md
Browse files
README.md
CHANGED
|
@@ -6,6 +6,11 @@ base_model: jondurbin/bagel-2.8b-v0.2
|
|
| 6 |
---
|
| 7 |
**NOTE**: These weights require the latest build of llama.cpp (i.e. commit [`c2101a2`](https://github.com/ggerganov/llama.cpp/commit/c2101a2e909ac7c08976d414e64e96c90ee5fa9e)). More details [here](https://github.com/ggerganov/llama.cpp/pull/5328).
|
| 8 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
From the author of the llama.cpp patch:
|
| 10 |
> I started working on this as an experiment and because I wanted to try Mamba models with llama.cpp (also, there have been quite a few finetunes already).
|
| 11 |
Turns out that implementing support for a novel model architecture is quite fun (well, at least when it finally works).
|
|
|
|
| 6 |
---
|
| 7 |
**NOTE**: These weights require the latest build of llama.cpp (i.e. commit [`c2101a2`](https://github.com/ggerganov/llama.cpp/commit/c2101a2e909ac7c08976d414e64e96c90ee5fa9e)). More details [here](https://github.com/ggerganov/llama.cpp/pull/5328).
|
| 8 |
|
| 9 |
+
How to use:
|
| 10 |
+
```
|
| 11 |
+
./main -e -p "[INST] Write an essay about AI. [/INST]\n" -m ggml-bagel-2.8b-v0.2-q8_0.gguf
|
| 12 |
+
```
|
| 13 |
+
|
| 14 |
From the author of the llama.cpp patch:
|
| 15 |
> I started working on this as an experiment and because I wanted to try Mamba models with llama.cpp (also, there have been quite a few finetunes already).
|
| 16 |
Turns out that implementing support for a novel model architecture is quite fun (well, at least when it finally works).
|