Update README.md
Browse files
README.md
CHANGED
@@ -24,6 +24,18 @@ Each branch contains an individual bits per weight, with the main one containing
|
|
24 |
|
25 |
Original model: https://huggingface.co/Locutusque/OpenCerebrum-2.0-7B
|
26 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
27 |
| Branch | Bits | lm_head bits | VRAM (4k) | VRAM (16k) | VRAM (32k) | Description |
|
28 |
| ----- | ---- | ------- | ------ | ------ | ------ | ------------ |
|
29 |
| [8_0](https://huggingface.co/bartowski/OpenCerebrum-2.0-7B-exl2/tree/8_0) | 8.0 | 8.0 | 8.4 GB | 9.8 GB | 11.8 GB | Maximum quality that ExLlamaV2 can produce, near unquantized performance. |
|
|
|
24 |
|
25 |
Original model: https://huggingface.co/Locutusque/OpenCerebrum-2.0-7B
|
26 |
|
27 |
+
## Prompt format
|
28 |
+
|
29 |
+
No chat template specified so ChatML is used. This may be incorrect, check original model card for details.
|
30 |
+
|
31 |
+
```
|
32 |
+
<|im_start|>system
|
33 |
+
{message}<|im_end|>
|
34 |
+
<|im_start|>user
|
35 |
+
{user message}<|im_end|>
|
36 |
+
<|im_start|>assistant
|
37 |
+
```
|
38 |
+
|
39 |
| Branch | Bits | lm_head bits | VRAM (4k) | VRAM (16k) | VRAM (32k) | Description |
|
40 |
| ----- | ---- | ------- | ------ | ------ | ------ | ------------ |
|
41 |
| [8_0](https://huggingface.co/bartowski/OpenCerebrum-2.0-7B-exl2/tree/8_0) | 8.0 | 8.0 | 8.4 GB | 9.8 GB | 11.8 GB | Maximum quality that ExLlamaV2 can produce, near unquantized performance. |
|