Update README.md

README.md

@@ -1,22 +1,130 @@
+---
+license: mit
+library_name: transformers
+base_model: unsloth/gpt-oss-20b
+tags:
+- gpt-oss
+- lora
+- unsloth
+- text-generation
+- instruction-following
+- multilingual
+datasets:
+- HuggingFaceH4/Multilingual-Thinking
+pipeline_tag: text-generation
+language:
+- en
+---
+
+# GPT-OSS-20B Fine-Tuned
+
+A fine-tuned **gpt-oss-20b** model optimized for *efficient text generation, multilingual conversational tasks, and instruction-following*.
+
+---
+
+## Overview
+
+| Item | Details |
+|---|---|
+| **Base checkpoint** | `unsloth/gpt-oss-20b` |
+| **Fine-tune method** | LoRA (PEFT) with Unsloth |
+| **Training run** | 30 steps • Multilingual-Thinking dataset |
+| **Trainable params** | [To be calculated, if available] |
+| **Loss** | [Loss metrics unavailable] |
+| **Hardware** | [Hardware details unavailable] |
+| **License** | MIT License (Base model: Refer to gpt-oss-20b license) |
+| **Intended use** | Educational, research, and chat-based applications |
+
+---
+
+## Datasets
+
+| Dataset | Size | Focus |
+|---|---|---|
+| `HuggingFaceH4/Multilingual-Thinking` | [Size unavailable] | Multilingual reasoning and conversational tasks |
+
+The dataset was wrapped with the **chat template** before training.
+
+---
+
+## Installation
+
+To use this model, install the required dependencies:
+
+```bash
+pip install torch>=2.8.0 triton>=3.4.0 transformers>=4.55.3 bitsandbytes unsloth
+```
+
+## Usage
+
+### Loading the Model
+
+```python
+from unsloth import FastLanguageModel
+import torch
+
+model, tokenizer = FastLanguageModel.from_pretrained(
+    model_name="unsloth/gpt-oss-20b",
+    max_seq_length=1024,
+    dtype=torch.float16,
+    load_in_4bit=True,
+)
+```
+
+### Fine-Tuning with LoRA
+
+```python
+model = FastLanguageModel.get_peft_model(
+    model,
+    r=8,
+    target_modules=["q_proj", "k_proj", "v_proj", "o_proj", "gate_proj", "up_proj", "down_proj"],
+    lora_alpha=16,
+    lora_dropout=0,
+    bias="none",
+    use_gradient_checkpointing="unsloth",
+)
+```
+
+### Inference
+
+```python
+from transformers import TextStreamer
+
+messages = [
+    {"role": "user", "content": "Solve x^5 + 3x^4 - 10 = 3."},
+]
+inputs = tokenizer.apply_chat_template(messages, return_tensors="pt", return_dict=True).to(model.device)
+
+outputs = model.generate(**inputs, max_new_tokens=512, streamer=TextStreamer(tokenizer))
+```
+
 ---
-base_model: unsloth/gpt-oss-20b-unsloth-bnb-4bit
-tags:
-- text-generation-inference
-- transformers
-- unsloth
-- gpt_oss
-- trl
-license: apache-2.0
-language:
-- en
+
+## Training Details
+
+### Training Configuration
+
+- **Batch Size**: 1
+- **Gradient Accumulation Steps**: 4
+- **Learning Rate**: 2e-4
+- **Optimizer**: adamw_8bit
+- **Warmup Steps**: 5
+- **Max Steps**: 30
+
 ---
 
-# Uploaded  model
+## Responsible Use
+
+- **Bias**: The model may reflect biases in the training data. Users should evaluate outputs for fairness.
+- **Misuse**: Avoid using for harmful or misleading content generation.
+- **Limitations**: Optimized for efficiency with 4-bit quantization, which may introduce minor accuracy trade-offs. Limited to 1024-token sequences.
+- **Disclaimer**: Not intended for critical decision-making. The author and base-model creators accept no liability for misuse or errors.
+
+---
 
-- **Developed by:** prxshetty
-- **License:** apache-2.0
-- **Finetuned from model :** unsloth/gpt-oss-20b-unsloth-bnb-4bit
+## Acknowledgements
 
-This gpt_oss model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
+- The unsloth library for enabling efficient fine-tuning.
+- Hugging Face for providing the base model and training infrastructure.
 
-[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
+---