---
library_name: transformers
extra_gated_heading: Access Gemma on Hugging Face
extra_gated_prompt: >-
  To access CodeGemma on Hugging Face, you’re required to review and agree to
  Google’s usage license. To do this, please ensure you’re logged-in to Hugging
  Face and click below. Requests are processed immediately.
extra_gated_button_content: Acknowledge license
pipeline_tag: text-generation
widget:
- text: >
    <start_of_turn>user
    Write a Python function to calculate the nth fibonacci number.<end_of_turn>
    <start_of_turn>model
inference:
  parameters:
    max_new_tokens: 200
license: gemma
license_link: https://ai.google.dev/gemma/terms
quantized_by: bartowski
---

## Exllama v2 Quantizations of codegemma-7b-it

Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.0.18">turboderp's ExLlamaV2 v0.0.18</a> for quantization.

<b>The "main" branch only contains the measurement.json, download one of the other branches for the model (see below)</b>

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/google/codegemma-7b-it

## Prompt format

```
<bos><start_of_turn>user
{prompt}<end_of_turn>
<start_of_turn>model

```

## Available sizes

No GQA - VRAM requirements will be higher

| Branch                                                         | Bits | lm_head bits | Size (4k) | Size (16k) | Description |
| -------------------------------------------------------------- | ---- | ------------ | --------- | ---------- | ----------- |
| [8_0](https://huggingface.co/bartowski/codegemma-7b-it-exl2/tree/8_0) | 8.0 | 8.0 | 14.0 GB | 19.4 GB | Maximum quality that ExLlamaV2 can produce, near unquantized performance. |
| [6_5](https://huggingface.co/bartowski/codegemma-7b-it-exl2/tree/6_5) | 6.5  | 8.0 | 12.5 GB | 17.9 GB | Near unquantized performance at vastly reduced size, **recommended**. |
| [5_0](https://huggingface.co/bartowski/codegemma-7b-it-exl2/tree/5_0) | 5.0  | 6.0 | 10.9 GB | 16.3 GB | Slightly lower quality vs 6.5, but usable on 8GB cards with 4k context. |
| [4_25](https://huggingface.co/bartowski/codegemma-7b-it-exl2/tree/4_25) | 4.25 | 6.0 | 10.2 GB | 15.7 GB | GPTQ equivalent bits per weight. |
| [3_5](https://huggingface.co/bartowski/codegemma-7b-it-exl2/tree/3_5) | 3.5 | 6.0 | 9.5 GB | 14.9 GB | Lower quality, not recommended. |

## Download instructions

With git:

```shell
git clone --single-branch --branch 6_5 https://huggingface.co/bartowski/codegemma-7b-it-exl2 codegemma-7b-it-exl2-6_5
```

With huggingface hub (credit to TheBloke for instructions):

```shell
pip3 install huggingface-hub
```

To download the `main` (only useful if you only care about measurement.json) branch to a folder called `codegemma-7b-it-exl2`:

```shell
mkdir codegemma-7b-it-exl2
huggingface-cli download bartowski/codegemma-7b-it-exl2 --local-dir codegemma-7b-it-exl2 --local-dir-use-symlinks False
```

To download from a different branch, add the `--revision` parameter:

Linux:

```shell
mkdir codegemma-7b-it-exl2-6_5
huggingface-cli download bartowski/codegemma-7b-it-exl2 --revision 6_5 --local-dir codegemma-7b-it-exl2-6_5 --local-dir-use-symlinks False
```

Windows (which apparently doesn't like _ in folders sometimes?):

```shell
mkdir codegemma-7b-it-exl2-6.5
huggingface-cli download bartowski/codegemma-7b-it-exl2 --revision 6_5 --local-dir codegemma-7b-it-exl2-6.5 --local-dir-use-symlinks False
```

Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski