Update README.md

README.md

@@ -1,78 +1,16 @@
----
-license: apache-2.0
-base_model: Qwen/Qwen3-4B-Instruct-2507
-tags:
-  - fitness
-  - pose-correction
-  - exercise-coaching
-  - fine-tuned
-language:
-  - en
-  - hu
----
-
-# PoseFit Correction Model
-
-Fine-tuned Qwen3-4B model for generating real-time exercise correction commands based on pose analysis.
-
-## Model Description
-
-This model analyzes exercise movement data (angles, phases, orientation) and provides short, actionable correction commands in English and Hungarian.
-
-**Base Model:** Qwen/Qwen3-4B-Instruct-2507  
-**Fine-tuning:** LoRA (merged)  
-**Training:** ~700 synthetic samples across 11 exercises
-
-## Supported Exercises
-
-- Calf Raises
-- Glute Bridges
-- Heel Touches
-- Jumping Jacks
-- Lunges
-- Mountain Climbers
-- Plank
-- Pushups
-- Russian Twist
-- Sit-ups
-- Squat
-
-## Input Format
-
-```
-Mean Actual Angle: 133.8 degrees
-Mean Target Angle: 171.6 degrees
-Joint: left_shoulder-left_hip-left_knee
-Phase: throughout
-Orientation: horizontal
-Error Category: insufficient_extension
-```
-
-## Example Output
-
-```
-Lower your hips, straight line.
-```
-
-## Usage
-
-```python
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-model = AutoModelForCausalLM.from_pretrained("roli253/posefit-correction")
-tokenizer = AutoTokenizer.from_pretrained("roli253/posefit-correction")
-
-messages = [
-    {"role": "system", "content": "You are a professional fitness coach. You analyze movement data and provide short, actionable correction commands."},
-    {"role": "user", "content": "Analyze the following movement analysis and provide a correction command in English.\n\nAnalysis:\nMean Actual Angle: 133.8 degrees\nMean Target Angle: 171.6 degrees\nJoint: left_shoulder-left_hip-left_knee\nPhase: throughout\nOrientation: horizontal\nError Category: insufficient_extension"}
-]
-
-text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
-inputs = tokenizer(text, return_tensors="pt")
-outputs = model.generate(**inputs, max_new_tokens=50)
-print(tokenizer.decode(outputs[0], skip_special_tokens=True))
-```
-
-## License
-
-Apache 2.0
+---
+license: apache-2.0
+base_model: Qwen/Qwen3-4B-Instruct-2507
+tags:
+  - fitness
+  - pose-correction
+  - exercise-coaching
+  - fine-tuned
+language:
+  - en
+  - hu
+---
+
+# PoseFit Correction Model
+
+Fine-tuned Qwen3-4B model for generating real-time exercise correction commands based on pose analysis.