initial commit

README.md

@@ -17,13 +17,8 @@ CVEParrot is a Google T5 model fine-tuned on CVE (Common Vulnerabilities and Exp
 
 ## Model Description
 
-- **Developed by:** findthehead
-- **Base Model:** Google T5
-- **Training Data:** CVE Database
-- **Language:** English
-- **License:** Apache 2.0
-
-This model has been specifically trained to understand and generate content related to cybersecurity vulnerabilities, CVE descriptions, and security intelligence.
+- **Developed by:** [Subhay Roy Chowdhury(findthehead)](https://huggingface.co/findthehead)
+- **Base Model:** Google T5 Small
 
 ## Use Cases
 
@@ -33,86 +28,39 @@ This model has been specifically trained to understand and generate content rela
 - Automated vulnerability documentation
 - CVE information extraction and summarization
 
-## How to Use
-
-### Option 1: Using Hugging Face Transformers (Safetensors)
-
-Install the required dependencies:
-
-```bash
-pip install transformers torch
-```
-
-**Inference Code:**
+## Inference Code
 
 ```python
-from transformers import T5Tokenizer, T5ForConditionalGeneration
-
-# Load model and tokenizer
+import warnings
+import os
+os.environ["TF_ENABLE_ONEDNN_OPTS"] = "0"
+os.environ["TF_CPP_MIN_LOG_LEVEL"] = "3"
+os.environ["TRANSFORMERS_NO_ADVISORY_WARNINGS"] = "1"
+warnings.filterwarnings("ignore")
+import torch
+from transformers import AutoTokenizer, T5ForConditionalGeneration
 model_name = "Prachir-AI/cveparrot"
-tokenizer = T5Tokenizer.from_pretrained(model_name)
-model = T5ForConditionalGeneration.from_pretrained(model_name)
-
-# Prepare input
-input_text = "Describe CVE-2024-1234"
-input_ids = tokenizer(input_text, return_tensors="pt").input_ids
-
-# Generate output
-outputs = model.generate(
-    input_ids,
-    max_length=512,
-    num_beams=4,
-    early_stopping=True,
-    temperature=0.7,
-    do_sample=True
-)
+tokenizer = AutoTokenizer.from_pretrained(model_name, use_fast=False)
 
-# Decode and print result
-generated_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
-print(generated_text)
-```
 
-**Advanced Usage with Custom Parameters:**
+model = T5ForConditionalGeneration.from_pretrained(model_name)
 
-```python
-from transformers import T5Tokenizer, T5ForConditionalGeneration
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+model.to(device)
 
-# Load model and tokenizer
-model_name = "Prachir-AI/cveparrot"
-tokenizer = T5Tokenizer.from_pretrained(model_name)
-model = T5ForConditionalGeneration.from_pretrained(model_name)
+prompt = "Provide detailed information about CVE-2021-3184."
+inputs = tokenizer(prompt, return_tensors="pt").to(device)
 
-# Move to GPU if available
-import torch
-device = "cuda" if torch.cuda.is_available() else "cpu"
-model = model.to(device)
-
-# Example prompts
-prompts = [
-    "Explain the security vulnerability:",
-    "Describe the CVE:",
-    "What is the impact of:",
-]
-
-input_text = prompts[0] + " CVE-2024-1234"
-input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to(device)
-
-# Generate with custom parameters
-outputs = model.generate(
-    input_ids,
-    max_length=256,
-    min_length=50,
-    num_beams=5,
-    no_repeat_ngram_size=2,
-    early_stopping=True,
-    temperature=0.8,
-    top_k=50,
-    top_p=0.95,
-    do_sample=True
-)
+with torch.no_grad():
+    output_ids = model.generate(
+        **inputs,
+        max_new_tokens=128,
+        temperature=1.0,
+        do_sample=True,
+    )
 
-generated_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
-print(generated_text)
+response = tokenizer.decode(output_ids[0], skip_special_tokens=True)
+print(response)
 ```
 
 ### Option 2: Using GGUF Model with Ollama (Local Inference)
@@ -208,50 +156,4 @@ This model was fine-tuned on CVE database entries to understand and generate sec
 - CVE descriptions and technical details
 - Vulnerability severity and impact analysis
 - Security patches and mitigation strategies
-- Affected software and version information
-
-## Limitations
-
-- The model is trained on historical CVE data and may not have information about very recent vulnerabilities
-- Generated content should be verified against official CVE databases
-- The model may occasionally generate plausible but incorrect security information
-- Not a replacement for professional security analysis
-
-## Ethical Considerations
-
-This model is designed for:
-- ✅ Security research and education
-- ✅ Vulnerability analysis and documentation
-- ✅ Automated security intelligence gathering
-- ✅ Assisting security professionals
-
-This model should NOT be used for:
-- ❌ Creating or exploiting vulnerabilities
-- ❌ Malicious hacking activities
-- ❌ Unauthorized security testing
-
-## Citation
-
-If you use this model in your research or applications, please cite:
-
-```bibtex
-@model{cveparrot2024,
-  author = {findthehead},
-  title = {CVEParrot: A T5 Model for CVE Analysis},
-  year = {2024},
-  publisher = {HuggingFace},
-  url = {https://huggingface.co/Prachir-AI/cveparrot}
-}
-```
-
-## Developer
-
-- **HuggingFace:** [findthehead](https://huggingface.co/findthehead)
-
-## Feedback and Contributions
-
-For issues, questions, or contributions, please visit the model repository on HuggingFace.
-
-## License
-
-This model is released under the Apache 2.0 License. See LICENSE file for details.
+- Affected software and version information