Add VRAM usage table

README.md

@@ -22,23 +22,19 @@ pipeline_tag: text-generation
 
 Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.0.11">turboderp's ExLlamaV2 v0.0.11</a> for quantization.
 
-Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions.
-
-Conversion was done using the default calibration dataset.
+<b>The "main" branch only contains the measurement.json, download one of the other branches for the model (see below)</b>
 
-Default arguments used except when the bits per weight is above 6.0, at that point the lm_head layer is quantized at 8 bits per weight instead of the default 6.
+Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions.
 
 Original model: https://huggingface.co/NousResearch/Nous-Hermes-2-SOLAR-10.7B
  
-<a href="https://huggingface.co/bartowski/Nous-Hermes-2-SOLAR-10.7B-exl2/tree/3_5">3.5 bits per weight</a>
- 
-<a href="https://huggingface.co/bartowski/Nous-Hermes-2-SOLAR-10.7B-exl2/tree/4_0">4.0 bits per weight</a>
- 
-<a href="https://huggingface.co/bartowski/Nous-Hermes-2-SOLAR-10.7B-exl2/tree/5_0">5.0 bits per weight</a>
- 
-<a href="https://huggingface.co/bartowski/Nous-Hermes-2-SOLAR-10.7B-exl2/tree/6_5">6.5 bits per weight</a>
- 
-<a href="https://huggingface.co/bartowski/Nous-Hermes-2-SOLAR-10.7B-exl2/tree/8_0">8.0 bits per weight</a>
+| Branch | Bits | lm_head bits | VRAM (4k) | VRAM (16k) | VRAM (32k) | Description |
+| ----- | ---- | ------- | ------ | ------ | ------ | ------------ |
+| [8_0](https://huggingface.co/bartowski/Nous-Hermes-2-SOLAR-10.7B-exl2/tree/8_0) | 8.0 | 8.0 | 11.9 GB | 13.3 GB | 15.3 GB | Maximum quality that ExLlamaV2 can produce, near unquantized performance. |
+| [6_5](https://huggingface.co/bartowski/Nous-Hermes-2-SOLAR-10.7B-exl2/tree/6_5) | 6.5 | 8.0 | 10.3 GB | 11.7 GB | 13.7 GB | Very similar to 8.0, good tradeoff of size vs performance, **recommended**. |
+| [5_0](https://huggingface.co/bartowski/Nous-Hermes-2-SOLAR-10.7B-exl2/tree/5_0) | 5.0 | 6.0 | 8.3 GB | 9.7 GB | 11.7 GB | Slightly lower quality vs 6.5, but usable on 8GB cards. |
+| [4_25](https://huggingface.co/bartowski/Nous-Hermes-2-SOLAR-10.7B-exl2/tree/4_25) | 4.25 | 6.0 | 7.4 GB | 8.6 GB | 10.6 GB | GPTQ equivalent bits per weight, slightly higher quality. |
+| [3_5](https://huggingface.co/bartowski/Nous-Hermes-2-SOLAR-10.7B-exl2/tree/3_5) | 3.5 | 6.0 | 6.4 GB | 7.8 GB | 9.8 GB | Lower quality, only use if you have to. |
 
 ## Download instructions