Update README.md
Browse files
README.md
CHANGED
|
@@ -20,11 +20,13 @@ The GGUFs uploaded are full FP32 precision.
|
|
| 20 |
Using OpenOrca GPT-4 data + cosmopedia for some extra data + dolly15k for instruct
|
| 21 |
|
| 22 |
## Model Details:
|
| 23 |
-
-
|
| 24 |
- 8 attention heads
|
| 25 |
-
- 32 layers
|
| 26 |
- 384 embeddings size
|
| 27 |
-
- 2048/8192 context (please use 4x RoPE scaling)
|
|
|
|
|
|
|
| 28 |
|
| 29 |
## Prompt Format (Alpaca):
|
| 30 |
```
|
|
|
|
| 20 |
Using OpenOrca GPT-4 data + cosmopedia for some extra data + dolly15k for instruct
|
| 21 |
|
| 22 |
## Model Details:
|
| 23 |
+
- 71.7M parameters (71,775,700)
|
| 24 |
- 8 attention heads
|
| 25 |
+
- 32 layers (34 layers on final model)
|
| 26 |
- 384 embeddings size
|
| 27 |
+
- 2048/8192/16384 context (please use 4x RoPE scaling, may train a 16k finetuned version later)
|
| 28 |
+
- Batch size 16
|
| 29 |
+
- llama.cpp (train-text-from-scratch)
|
| 30 |
|
| 31 |
## Prompt Format (Alpaca):
|
| 32 |
```
|