Update README.md

README.md

@@ -2,27 +2,49 @@
 language:
 - en
 license: other
-base_model: unsloth/Phi-4
-library_name: peft
+base_model: microsoft/phi-2  # Update with your base model
+library_name: peft transformers bitsandbytes
 tags:
 - text-generation
-- peft
 - lora
-- quantization
-model_type: phi-4b-peft
+- 4bit-quantization
+- autopeft
+model_type: causal-lm
 inference:
   parameters:
-    temperature: 0
-    min_p: 0.95
+    temperature: 0.0
+    min_p: 1.0
     max_new_tokens: 2048
 pipeline_tag: text-generation
+task_type: CAUSAL_LM
 quantization_config:
   load_in_4bit: true
   bnb_4bit_compute_dtype: float16
   bnb_4bit_quant_type: nf4
-  llm_int8_threshold: 6
-  llm_int8_skip_modules: null
-  llm_int8_enable_fp32_cpu_offload: false
-  llm_int8_has_fp16_weight: false
-task_type: CAUSAL_LM
----
+  bnb_4bit_use_double_quant: true
+---
+
+# Model Card for Phi-4 LoRA Model
+
+## Model Description
+
+Quantized LoRA adapter for Microsoft's Phi-2 model, fine-tuned for light pattern generation tasks.
+
+## Usage
+
+```python
+from peft import AutoPeftModelForCausalLM
+from transformers import AutoTokenizer
+
+# Load model with 4-bit quantization
+model = AutoPeftModelForCausalLM.from_pretrained(
+    "your-username/phi4_lora_model",
+    load_in_4bit=True,  # Automatically uses config from quantization_config
+    device_map="auto"
+)
+
+# Load standard Transformers tokenizer
+tokenizer = AutoTokenizer.from_pretrained(
+    "your-username/phi4_lora_model",
+    use_fast=True
+)