Update README.md
Browse files
README.md
CHANGED
|
@@ -21,6 +21,13 @@ Another experimental model, using mostly sythetic data generated by [airoboros](
|
|
| 21 |
|
| 22 |
This fine-tune is on [jamba-v0.1](https://huggingface.co/ai21labs/Jamba-v0.1) using QLoRA.
|
| 23 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
#### Highlights
|
| 25 |
|
| 26 |
The base model, jamba-v0.1, supposedly has a 256k context length, but this finetune was done with 8192 so anything beyond that will likely have undefined results.
|
|
|
|
| 21 |
|
| 22 |
This fine-tune is on [jamba-v0.1](https://huggingface.co/ai21labs/Jamba-v0.1) using QLoRA.
|
| 23 |
|
| 24 |
+
Be sure to install the dependencies for jamba:
|
| 25 |
+
|
| 26 |
+
```bash
|
| 27 |
+
pip install transformers>=4.39.0
|
| 28 |
+
pip install mamba-ssm causal-conv1d>=1.2.0
|
| 29 |
+
```
|
| 30 |
+
|
| 31 |
#### Highlights
|
| 32 |
|
| 33 |
The base model, jamba-v0.1, supposedly has a 256k context length, but this finetune was done with 8192 so anything beyond that will likely have undefined results.
|