README.md

---
language:
- multilingual
license: gemma
library_name: transformers
tags:
- nlp
- code
base_model: google/gemma-2-2b-jpn-it
datasets:
- mlabonne/orpo-dpo-mix-40k
license_link: https://ai.google.dev/gemma/terms
pipeline_tag: text-generation
quantized_by: ymcki
widget:
- messages:
  - role: user
    content: Can you provide ways to eat combinations of bananas and dragonfruits?
model-index:
- name: gemma-2-2b-jpn-it-abliterated-17-ORPO
  results:
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: IFEval (0-Shot)
      type: HuggingFaceH4/ifeval
      args:
        num_few_shot: 0
    metrics:
    - type: inst_level_strict_acc and prompt_level_strict_acc
      value: 49.48
      name: strict accuracy
    source:
      url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ymcki/gemma-2-2b-jpn-it-abliterated-17-ORPO
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: BBH (3-Shot)
      type: BBH
      args:
        num_few_shot: 3
    metrics:
    - type: acc_norm
      value: 14.92
      name: normalized accuracy
    source:
      url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ymcki/gemma-2-2b-jpn-it-abliterated-17-ORPO
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: MATH Lvl 5 (4-Shot)
      type: hendrycks/competition_math
      args:
        num_few_shot: 4
    metrics:
    - type: exact_match
      value: 2.87
      name: exact match
    source:
      url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ymcki/gemma-2-2b-jpn-it-abliterated-17-ORPO
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: GPQA (0-shot)
      type: Idavidrein/gpqa
      args:
        num_few_shot: 0
    metrics:
    - type: acc_norm
      value: 3.24
      name: acc_norm
    source:
      url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ymcki/gemma-2-2b-jpn-it-abliterated-17-ORPO
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: MuSR (0-shot)
      type: TAUR-Lab/MuSR
      args:
        num_few_shot: 0
    metrics:
    - type: acc_norm
      value: 5.67
      name: acc_norm
    source:
      url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ymcki/gemma-2-2b-jpn-it-abliterated-17-ORPO
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: MMLU-PRO (5-shot)
      type: TIGER-Lab/MMLU-Pro
      config: main
      split: test
      args:
        num_few_shot: 5
    metrics:
    - type: acc
      value: 13.18
      name: accuracy
    source:
      url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ymcki/gemma-2-2b-jpn-it-abliterated-17-ORPO
      name: Open LLM Leaderboard
---

Original model: https://huggingface.co/google/gemma-2-2b-jpn-it

## Prompt format

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

```

Note that this model does not support a System prompt.

This is abliterated model of [google/gemma-2-2b-jpn-it](https://huggingface.co/google/gemma-2-2b-jpn-it) using the 
[method](https://medium.com/@mlabonne/uncensor-any-llm-with-abliteration-d30148b7d43e) 
described by mlabonne.

Layer 17 of the original model was chosen for abliteration.
I also created another layer 18 abliterated model for comparison.

ORPO fine tuning was performed for four epoches.

| Epoch | loss | eval_loss |
| ----- | ---- | --------- |
| 1 | 1.20152769684791564 | 1.0501047372817993 |
| 2 | 1.25755584239959716 | 1.0144596099853516 |
| 3 | 0.93099724054336543 | 0.9957754611968994 |
| 4 | 0.88664623498916623 | 0.9857067465782166 |

The fine tuned model is uploaded here to be evaluated by the Open LLM Leaderboard to see if the slightly brain damaged non-ORPO model can be healed. Again, the fine tuning method is also based on one described by [mlabonne](https://towardsdatascience.com/fine-tune-llama-3-with-orpo-56cfab2f9ada) but the input model was read into VRAM by [unsloth](https://github.com/unslothai/unsloth) to allow using the full 40k dataset to run on a single 3090.

## Benchmark (100.0*raw scores only)

Click on the model name go to the raw score json generated by Open LLM Leaderboard.

| Model | Average | IFEval | BHH | Math Lv5 | GPQA | MUSR | MMLU-PRO |
| ----- | ------- | ------ | ----|--------- | ---- | ---- | -------- |
| [gemma-2-2b-jpn-it](https://huggingface.co/datasets/open-llm-leaderboard/results/blob/main/google/gemma-2-2b-jpn-it/results_2024-10-15T15-21-39.173019.json) | 30.82 | 54.11 | 41.43 | 0.0 | 27.52 | 37.17 | 24.67 |
| [gemma-2-2b-jpn-it-abliterated-17-ORPO](https://huggingface.co/datasets/open-llm-leaderboard/results/raw/main/ymcki/gemma-2-2b-jpn-it-abliterated-17-ORPO/results_2024-10-20T02-46-59.069357.json) | 29.99 | 50.94 | 38.59 | 2.87 | 27.43 | 38.23 | 21.86 |
| [gemma-2-2b-jpn-it-abliterated-17](https://huggingface.co/datasets/open-llm-leaderboard/results/raw/main/ymcki/gemma-2-2b-jpn-it-abliterated-17/results_2024-10-18T15-18-46.821674.json) | 30.29 | 52.65 | 40.46 | 0.0 | 27.18 | 36.90 | 24.55 |
| [gemma-2-2b-jpn-it-abliterated-18](https://huggingface.co/datasets/open-llm-leaderboard/results/raw/main/ymcki/gemma-2-2b-jpn-it-abliterated-18/results_2024-10-18T15-41-42.399571.json) | 30.61 | 53.02 | 40.96 | 0.0 | 27.35 | 37.30 | 25.05 |

Looks like fine tuning is probably not enough. May need to run more epoches.

## How to run this model

```py
from transformers import AutoTokenizer, AutoModelForCausalLM
import transformers
import torch

model_id = "gemma-2-2b-jpn-it-abliterated-17-ORPO"
dtype = torch.bfloat16

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    device_map="cuda",
    torch_dtype=dtype,)

chat = [
    { "role": "user", "content": "Write a hello world program" },
]
prompt = tokenizer.apply_chat_template(chat, tokenize=False, add_generation_prompt=True)
```

## Downloading using huggingface-cli

First, make sure you have hugginface-cli installed:

```
pip install -U "huggingface_hub[cli]"
```

Then, you can target the specific file you want:

```
huggingface-cli download ymcki/gemma-2-2b-jpn-it-abliterated-17-ORPO --include "*" --local-dir ./
```

## Credits

Thank you mlabonne for describing his fine tuning method.

Thanks FullOf_Bad_Ideas from LocalLlama for the suggestion of using unsloth to save VRAM.

# [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard)
Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_ymcki__gemma-2-2b-jpn-it-abliterated-17-ORPO)

|      Metric       |Value|
|-------------------|----:|
|Avg.               |14.89|
|IFEval (0-Shot)    |49.48|
|BBH (3-Shot)       |14.92|
|MATH Lvl 5 (4-Shot)| 2.87|
|GPQA (0-shot)      | 3.24|
|MuSR (0-shot)      | 5.67|
|MMLU-PRO (5-shot)  |13.18|