Update VRAM
Browse files
README.md
CHANGED
@@ -37,11 +37,11 @@ No GQA - VRAM requirements will be higher
|
|
37 |
|
38 |
| Branch | Bits | lm_head bits | Size (4k) | Size (16k) | Description |
|
39 |
| -------------------------------------------------------------- | ---- | ------------ | --------- | ---------- | ----------- |
|
40 |
-
| [8_0](https://huggingface.co/bartowski/gemma-7b-openhermes-exl2/tree/8_0) | 8.0 | 8.0 |
|
41 |
-
| [6_5](https://huggingface.co/bartowski/gemma-7b-openhermes-exl2/tree/6_5) | 6.5 | 8.0 |
|
42 |
-
| [5_0](https://huggingface.co/bartowski/gemma-7b-openhermes-exl2/tree/5_0) | 5.0 | 6.0 |
|
43 |
-
| [4_25](https://huggingface.co/bartowski/gemma-7b-openhermes-exl2/tree/4_25) | 4.25 | 6.0 |
|
44 |
-
| [3_5](https://huggingface.co/bartowski/gemma-7b-openhermes-exl2/tree/3_5) | 3.5 | 6.0 | 5
|
45 |
|
46 |
## Download instructions
|
47 |
|
|
|
37 |
|
38 |
| Branch | Bits | lm_head bits | Size (4k) | Size (16k) | Description |
|
39 |
| -------------------------------------------------------------- | ---- | ------------ | --------- | ---------- | ----------- |
|
40 |
+
| [8_0](https://huggingface.co/bartowski/gemma-7b-openhermes-exl2/tree/8_0) | 8.0 | 8.0 | 14.0 GB | 19.4 GB | Maximum quality that ExLlamaV2 can produce, near unquantized performance. |
|
41 |
+
| [6_5](https://huggingface.co/bartowski/gemma-7b-openhermes-exl2/tree/6_5) | 6.5 | 8.0 | 12.5 GB | 17.9 GB | Near unquantized performance at vastly reduced size, **recommended**. |
|
42 |
+
| [5_0](https://huggingface.co/bartowski/gemma-7b-openhermes-exl2/tree/5_0) | 5.0 | 6.0 | 10.9 GB | 16.3 GB | Slightly lower quality vs 6.5, great for 12GB cards with 4k context. |
|
43 |
+
| [4_25](https://huggingface.co/bartowski/gemma-7b-openhermes-exl2/tree/4_25) | 4.25 | 6.0 | 10.2 GB | 15.7 GB | GPTQ equivalent bits per weight, ideal for 16GB cards at 16k context |
|
44 |
+
| [3_5](https://huggingface.co/bartowski/gemma-7b-openhermes-exl2/tree/3_5) | 3.5 | 6.0 | 9.5 GB | 14.9 GB | Lower quality, not recommended. |
|
45 |
|
46 |
## Download instructions
|
47 |
|