|
|
--- |
|
|
license: apache-2.0 |
|
|
datasets: |
|
|
- prometheus-eval/Feedback-Collection |
|
|
language: |
|
|
- en |
|
|
base_model: |
|
|
- unsloth/gemma-3-4b-it |
|
|
pipeline_tag: image-text-to-text |
|
|
--- |
|
|
|
|
|
## Model Overview |
|
|
This model is a fine-tuned version of the `unsloth/gemma-3-4b-it` |
|
|
trained on the **Feedback-Collection** dataset from the *Prometheus Eval*. |
|
|
**Fine-tuning Framework:** Finetuned using Unsloth optimized LoRA adapters. |
|
|
|
|
|
--- |
|
|
|
|
|
## 🧮 Performance Benchmark |
|
|
|
|
|
| Model | Benchmark | Pearson r | Spearman ρ | |
|
|
|:------|:-----------|:----------:|:------------:| |
|
|
| 🟩 **This model** | Feedback Bench | **0.9198** | **0.9210** | |
|
|
| 🟨 **Prometheus 2 (8×7B)** *(Kim et al., 2024)* | Feedback Bench / Preference Bench | ≈ 0.898 / – | ≈ 0.90 / – | |
|
|
|
|
|
|
|
|
**Highlights:** |
|
|
- Achieves a better score on *Feedback Bench* (+0.02 higher than Prometheus 2). |
|
|
- Uses a **4B parameter model**, making it significantly lighter than Prometheus 2. |
|
|
- Demonstrates strong **semantic consistency and evaluative precision**. |
|
|
--- |
|
|
|
|
|
## 🧾 License |
|
|
|
|
|
This model is released under the **Apache 2.0 License**. |
|
|
However, because it is derived from **Google’s Gemma 3**, your use of this model must also comply with the **[Gemma Terms of Use](https://ai.google.dev/gemma/terms)**. |
|
|
|
|
|
By using this model, you agree to: |
|
|
- Follow Google’s **Gemma Model Terms of Use**, including restrictions on misuse and redistribution. |
|
|
- Attribute Google as the original provider of the Gemma 3 base model. |
|
|
|
|
|
For full details, see: [https://ai.google.dev/gemma/terms](https://ai.google.dev/gemma/terms) |
|
|
|
|
|
--- |
|
|
|
