Commit
•
2069da8
1
Parent(s):
1ea9dde
Update README.md
Browse files
README.md
CHANGED
|
@@ -51,7 +51,7 @@ The model was trained with flash attention and gradient checkpointing and deepsp
|
|
| 51 |
- Batch size: 128
|
| 52 |
- Max Length: 2048
|
| 53 |
- Learning rate: 5e-5
|
| 54 |
-
- Lora _r_:
|
| 55 |
- Lora Alpha: 32
|
| 56 |
|
| 57 |
## Prompting
|
|
@@ -80,7 +80,7 @@ from transformers import GenerationConfig
|
|
| 80 |
|
| 81 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
| 82 |
dtype = torch.float16
|
| 83 |
-
repo_id = "jordiclive/alpaca_gpt4-dolly_15k-vicuna-
|
| 84 |
base_model = "decapoda-research/llama-30b-hf"
|
| 85 |
|
| 86 |
# Model Loading
|
|
|
|
| 51 |
- Batch size: 128
|
| 52 |
- Max Length: 2048
|
| 53 |
- Learning rate: 5e-5
|
| 54 |
+
- Lora _r_: 64
|
| 55 |
- Lora Alpha: 32
|
| 56 |
|
| 57 |
## Prompting
|
|
|
|
| 80 |
|
| 81 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
| 82 |
dtype = torch.float16
|
| 83 |
+
repo_id = "jordiclive/lora-llama-33B-alpaca_gpt4-dolly_15k-vicuna-r64"
|
| 84 |
base_model = "decapoda-research/llama-30b-hf"
|
| 85 |
|
| 86 |
# Model Loading
|