Commit
•
a0b44c5
1
Parent(s):
4bb55b5
Update README.md
Browse filesUpdated with 5/3 exl2 quants with improved weights
README.md
CHANGED
|
@@ -39,6 +39,13 @@ Exl2 is available on Bullerwins's huggingface account. Check it out here:
|
|
| 39 |
[6.0bpw exl2](https://huggingface.co/bullerwins/gradientai_Llama-3-8B-Instruct-262k_exl2_6.0bpw)
|
| 40 |
[5.0bpw exl2](https://huggingface.co/bullerwins/gradientai_Llama-3-8B-Instruct-262k_exl2_5.0bpw)
|
| 41 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 42 |
**Data:**
|
| 43 |
|
| 44 |
For training data, we generate long contexts by augmenting [SlimPajama](https://huggingface.co/datasets/cerebras/SlimPajama-627B). We also fine-tune on a chat dataset based on UltraChat [4], following a similar recipe for data augmentation to [2].
|
|
|
|
| 39 |
[6.0bpw exl2](https://huggingface.co/bullerwins/gradientai_Llama-3-8B-Instruct-262k_exl2_6.0bpw)
|
| 40 |
[5.0bpw exl2](https://huggingface.co/bullerwins/gradientai_Llama-3-8B-Instruct-262k_exl2_5.0bpw)
|
| 41 |
|
| 42 |
+
**Updated Exl2 quants for 5/3 improved weights**
|
| 43 |
+
|
| 44 |
+
[8.0bpw exl2](https://huggingface.co/bullerwins/gradientai_Llama-3-8B-Instruct-262k_v2_exl2_8.0bpw)
|
| 45 |
+
[6.0bpw exl2](https://huggingface.co/bullerwins/gradientai_Llama-3-8B-Instruct-262k_v2_exl2_6.0bpw)
|
| 46 |
+
[5.0bpw exl2](https://huggingface.co/bullerwins/gradientai_Llama-3-8B-Instruct-262k_v2_exl2_5.0bpw)
|
| 47 |
+
|
| 48 |
+
|
| 49 |
**Data:**
|
| 50 |
|
| 51 |
For training data, we generate long contexts by augmenting [SlimPajama](https://huggingface.co/datasets/cerebras/SlimPajama-627B). We also fine-tune on a chat dataset based on UltraChat [4], following a similar recipe for data augmentation to [2].
|