Commit
•
ded9715
1
Parent(s):
7560135
Update README.md
Browse filesAdding a bit of extra info for clarification on what the attached dataset is for.
README.md
CHANGED
|
@@ -16,6 +16,8 @@ language:
|
|
| 16 |
This is a prune of [Llama 3 Refueled](https://www.huggingface.co/refuelai/llama-3-refueled) using [mergekit](https://github.com/cg123/mergekit) and [PruneMe](https://www.github.com/arcee-ai/PruneMe)
|
| 17 |
The model is semi-tested, but still needs some debugging, namely with converting to GGUF, though I am working on that.
|
| 18 |
|
|
|
|
|
|
|
| 19 |
### Performance
|
| 20 |
After only 1 test because of lack of compute and for stupid long inference times on my 3060ti (8GB), it does show some interesting results.
|
| 21 |
Here's the response after being prompted "Hi!" using the [example from Meta](https://llama.meta.com/docs/model-cards-and-prompt-formats/meta-llama-3).
|
|
|
|
| 16 |
This is a prune of [Llama 3 Refueled](https://www.huggingface.co/refuelai/llama-3-refueled) using [mergekit](https://github.com/cg123/mergekit) and [PruneMe](https://www.github.com/arcee-ai/PruneMe)
|
| 17 |
The model is semi-tested, but still needs some debugging, namely with converting to GGUF, though I am working on that.
|
| 18 |
|
| 19 |
+
Note: the [dataset](https://www.huggingface.co/yahma/alpaca-cleaned) was used for evaluating what layers should be pruned. This model was **NOT** (yet) finetuned.
|
| 20 |
+
|
| 21 |
### Performance
|
| 22 |
After only 1 test because of lack of compute and for stupid long inference times on my 3060ti (8GB), it does show some interesting results.
|
| 23 |
Here's the response after being prompted "Hi!" using the [example from Meta](https://llama.meta.com/docs/model-cards-and-prompt-formats/meta-llama-3).
|