Update README.md

README.md

@@ -48,87 +48,49 @@ Kalıp koruma sistemlerinden gelen log kayıtlarını analiz ederek:
 
 ## 🚀 Hızlı Başlangıç
 
-### Kurulum
+### Google Colabda Kurulum 
 
 ```bash
-pip install torch transformers peft accelerate bitsandbytes
+%%capture
+!pip install unsloth
+
+# Also get the latest nightly Unsloth!
+
+!pip uninstall unsloth -y && pip install --upgrade --no-cache-dir --no-deps git+https://github.com/unslothai/unsloth.git@nightly git+https://github.com/unslothai/unsloth-zoo.git
 ```
 
 ### Kullanım
 
 ```python
-from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig
-from peft import PeftModel
+from unsloth import FastLanguageModel
 import torch
-
-# Quantization konfigürasyonu
-quantization_config = BitsAndBytesConfig(
-    load_in_4bit=True,
-    bnb_4bit_compute_dtype=torch.float16,
-    bnb_4bit_quant_type="nf4",
-    bnb_4bit_use_double_quant=True,
-)
-
-# Model ve tokenizer yükleme
-model_name = "your-username/llama-8b-mold-protection"
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-
-# Pad token ayarlama
-if tokenizer.pad_token is None:
-    tokenizer.pad_token = tokenizer.eos_token
-
-model = AutoModelForCausalLM.from_pretrained(
-    model_name,
-    torch_dtype=torch.float16,
-    device_map="auto",
-    quantization_config=quantization_config,
-    trust_remote_code=True
+max_seq_length = 2048 # Choose any! We auto support RoPE Scaling internally!
+dtype = None # None for auto detection. Float16 for Tesla T4, V100, Bfloat16 for Ampere+
+load_in_4bit = True # Use 4bit quantization to reduce memory usage. Can be False.
+
+model, tokenizer = FastLanguageModel.from_pretrained(
+    model_name = "bbayrm0/lora_model",
+    max_seq_length = max_seq_length,
+    dtype = dtype,
+    load_in_4bit = load_in_4bit
 )
 
-# Log analizi örneği
-def analyze_log(log_entry):
-    prompt = f"""Kalıp koruma sistemi log girişini analiz et ve detaylı açıklama ile çözüm önerisi sun:
-
-{log_entry}
-
-Açıklama ve Çözüm:"""
-    
-    # Input'ları modelle aynı cihaza taşı
-    inputs = tokenizer(
-        prompt, 
-        return_tensors="pt", 
-        padding=True, 
-        truncation=True, 
-        max_length=512
-    )
-    
-    # Input tensor'ları GPU'ya taşı
-    inputs = {k: v.to(model.device) for k, v in inputs.items()}
-    
-    with torch.no_grad():
-        outputs = model.generate(
-            **inputs,
-            max_new_tokens=256,  # max_length yerine max_new_tokens kullan
-            temperature=0.7,
-            do_sample=True,
-            top_p=0.9,
-            top_k=50,
-            repetition_penalty=1.1,
-            pad_token_id=tokenizer.eos_token_id,
-            eos_token_id=tokenizer.eos_token_id,
-            early_stopping=True
-        )
-    
-    # Sadece yeni generate edilen kısmı al
-    generated_tokens = outputs[0][inputs['input_ids'].shape[1]:]
-    response = tokenizer.decode(generated_tokens, skip_special_tokens=True)
-    
-    return response.strip()
-
-# Örnek kullanım
-log_sample = "2025-01-15 14:30:22 | MoldProtection | CRITICAL | KALIP KORUMA UYARISI - Hatalı ROI'ler: ROI 1, ROI 2 | Tetikleyici: plc"
-result = analyze_log(log_sample)
-print(result)
+FastLanguageModel.for_inference(model) # Enable native 2x faster inference
+
+messages = [
+    {"role": "user", "content": "2025-09-01 11:25:55 | MoldProtection | CRITICAL | KALIP KORUMA UYARISI - Hatalı ROI'ler: ROI 2, ROI 3, ROI 4 | Tetikleyici: manual"},
+]
+inputs = tokenizer.apply_chat_template(
+    messages,
+    tokenize = True,
+    add_generation_prompt = True, # Must add for generation
+    return_tensors = "pt",
+).to("cuda")
+
+from transformers import TextStreamer
+text_streamer = TextStreamer(tokenizer, skip_prompt = True)
+_ = model.generate(input_ids = inputs, streamer = text_streamer, max_new_tokens = 250,
+                   use_cache = True, temperature = 1, min_p = 0.1)
 ```
 
 ## 📈 Training Detayları