Update README.md

README.md

@@ -27,152 +27,166 @@ widget:
 language:
 - en
 license: apache-2.0
-base_model: meta-llama/Meta-Llama-3-8B-Instruct
+base_model: unsloth/llama-3-8b-Instruct-bnb-4bit
 tags:
+- text-generation
 - llama3
-- text-rewriting
-- ai-humanizer
-- writing
-- lora
-- merged
+- rewriting
+- humanizer
+- ai-detection
 - gguf
-- transformers
+- editorial
+- instruction-tuned
 pipeline_tag: text-generation
 library_name: transformers
+model-index:
+- name: HumanFlow-Llama3-8B
+  results:
+  - task:
+      type: text-generation
+    dataset:
+      name: Internal Evaluation Suite
+      type: custom
+    metrics:
+    - type: BERTScore F1
+      value: 0.8424
+    - type: ROUGE-L
+      value: 0.0908
+    - type: Perplexity
+      value: 1.5242
+    - type: Text Overlap
+      value: 0.0528
 ---
 
-<p align="center">
-  <img src="assets/humanflow_logofinal.png" width="340"/>
-</p>
+<div align="center">
 
-<h1 align="center">HumanFlow-Llama3-8B</h1>
+<img src="https://huggingface.co/randhir302/HumanFlow/resolve/main/humanflow_logofinal.png" width="300"/>
 
-<p align="center"><strong>The antidote to robotic AI text.</strong></p>
+# HumanFlow-Llama3-8B
 
-<p align="center">
-Rewrite synthetic AI writing into language that feels natural, fluent, and genuinely human.
-</p>
+### Premium Open-Source AI Rewriting Model  
+### Converts Robotic AI Text into Natural Human Flow
 
-<p align="center">
+<p>
+<a href="https://huggingface.co/randhir302/HumanFlow">
+<img src="https://img.shields.io/badge/🤗%20HuggingFace-Live-orange">
+</a>
 
-![License](https://img.shields.io/badge/License-Apache%202.0-black?style=for-the-badge)
-![Base](https://img.shields.io/badge/Base-Llama--3%208B-black?style=for-the-badge)
-![Weights](https://img.shields.io/badge/Weights-Merged-black?style=for-the-badge)
-![Ready](https://img.shields.io/badge/GGUF-Ready-black?style=for-the-badge)
+<a href="https://github.com/iamhero2709/HumanFlow">
+<img src="https://img.shields.io/badge/GitHub-Repository-black">
+</a>
 
-</p>
+<a href="#">
+<img src="https://img.shields.io/badge/License-Apache--2.0-blue">
+</a>
 
----
+<a href="#">
+<img src="https://img.shields.io/badge/Model-Llama3--8B-green">
+</a>
+</p>
 
-## Why HumanFlow Exists
+**Built for creators, students, agencies, founders, and developers who need text that sounds genuinely human.**
 
-Most AI-generated writing is easy to spot.
+</div>
 
-It often sounds stiff, repetitive, emotionally flat, and structurally predictable. That hurts trust, lowers engagement, and makes content feel artificial.
+---
 
-**HumanFlow-Llama3-8B** was built to solve this problem.
+# Why HumanFlow?
 
-It transforms robotic AI text into writing that reads naturally, flows smoothly, and sounds human—while preserving the original meaning.
+Most AI outputs follow predictable patterns:
 
----
+- Over-polished wording  
+- Repetitive sentence rhythm  
+- Detectable structure  
+- Robotic transitions  
+- Generic emotional tone  
 
-## Quick Stats
+HumanFlow was fine-tuned to rewrite **structure, rhythm, sentence flow, and human texture** instead of only replacing words.
 
-| Metric | Value |
-|---|---|
-| Human-like Score | 99% |
-| Base Model | Llama-3 8B |
-| Fine-Tuning | Unsloth + LoRA |
-| Release Type | Fully Merged Weights |
-| Quantization | GGUF Ready |
-| License | Apache-2.0 |
+That means output feels written by a real person — not generated by a template.
 
 ---
 
-## Model Details
+# Performance Snapshot
 
-| Category | Value |
-|---|---|
-| Model Name | HumanFlow-Llama3-8B |
-| Architecture | Transformer Decoder |
-| Base Checkpoint | Meta-Llama-3-8B-Instruct |
-| Training Method | Supervised Fine-Tuning |
-| Adapter Method | LoRA |
-| Merge Status | Fully Merged |
-| Primary Use | AI Text Humanization |
-| Inference | Transformers / vLLM / GGUF |
+| Metric | Base Llama-3 8B | HumanFlow-Llama3-8B |
+|-------|----------------|---------------------|
+| GPTZero Human Score | 18% | **99%** |
+| Turnitin Human Score | 10% | **92%+** |
+| Readability | Generic | **Natural** |
+| Rewrite Quality | Basic | **Advanced** |
+| Human Tone | Low | **High** |
+
+> Measured on internal benchmark prompts using recommended generation settings.
 
 ---
 
-## Before vs After
+# Core Evaluation (Real Metrics)
 
-### Example 1 — Marketing Copy
+| Metric | Score | Meaning |
+|------|------|---------|
+| BERTScore F1 | **0.8424** | Strong semantic preservation |
+| ROUGE-L | **0.0908** | Lower overlap = true rewriting |
+| Perplexity | **1.5242** | High fluency |
+| Text Overlap | **0.0528** | Minimal copy behavior |
 
-**Input**  
-Our platform provides innovative solutions that maximize productivity and efficiency.
+---
 
-**Output**  
-Our platform helps people work faster and smarter with tools that make everyday tasks easier.
+# What Makes HumanFlow Different?
 
----
+## Structural Humanization
 
-### Example 2 — Student Writing
+Changes how ideas are expressed, not just vocabulary.
 
-**Input**  
-Climate change is a significant issue requiring immediate international cooperation.
+## Detector-Aware Fine-Tuning
 
-**Output**  
-Climate change is a serious global challenge, and solving it will require countries to act together now.
+Optimized toward natural writing signatures.
 
----
+## Meaning Preservation
 
-### Example 3 — Email Rewrite
+Keeps original intent while improving realism.
 
-**Input**  
-I am writing this email to inform you that the requested document has been attached.
+## Ready for Real Use
 
-**Output**  
-Just letting you know I’ve attached the document you requested.
+Use locally, integrate into apps, deploy in workflows.
 
 ---
 
-## Built For
+# Ideal Use Cases
 
-- SEO agencies improving readability  
-- Content teams polishing drafts  
-- Students improving natural tone  
-- Developers integrating rewrite pipelines  
-- Startup founders refining copy  
-- Chatbots with better responses  
-- Email automation workflows  
-- Social media content teams  
+- SEO Content Teams  
+- Blog Writers  
+- Agencies  
+- Students  
+- Researchers  
+- SaaS Products  
+- Email Personalization Tools  
+- AI Content Pipelines  
 
 ---
 
-## Evaluation
+# Before vs After
 
-Automated benchmark results measured at **temperature = 0.7**
+## Input
 
-| Metric | Value | Interpretation |
-|--------|-------|----------------|
-| BERTScore F1 | 0.8424 | Strong semantic similarity while rewriting |
-| ROUGE-L | 0.0908 | Low overlap indicates fresh generation |
-| Perplexity | 1.5242 | High fluency and coherence |
-| Text Overlap | 0.0528 | Minimal lexical copying |
+> In today’s rapidly evolving digital landscape, it is imperative for organizations to leverage strategic methodologies in order to maximize engagement.
 
-These results indicate HumanFlow preserves meaning while actively regenerating more natural phrasing.
+## HumanFlow Output
+
+> Online markets move fast. If a company wants attention, it needs smart strategy, clear messaging, and content people actually care about.
 
 ---
 
-## Usage
+# Quickstart (Transformers)
 
 ```python
 from transformers import AutoTokenizer, AutoModelForCausalLM
 import torch
 
-model_id = "randhir302/HumanFlow-Llama3-8B"
+model_id = "randhir302/HumanFlow"
 
 tokenizer = AutoTokenizer.from_pretrained(model_id)
+
 model = AutoModelForCausalLM.from_pretrained(
     model_id,
     torch_dtype=torch.float16,
@@ -180,18 +194,19 @@ model = AutoModelForCausalLM.from_pretrained(
 )
 
 prompt = """
-Rewrite this text naturally:
+Rewrite this AI draft into natural human writing without changing meaning:
 
-The company offers innovative solutions that optimize workflow efficiency.
+Artificial intelligence is transforming industries worldwide.
 """
 
-inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
 
 outputs = model.generate(
     **inputs,
     max_new_tokens=220,
-    temperature=0.7,
-    top_p=0.9
+    temperature=0.75,
+    top_p=0.9,
+    repetition_penalty=1.1
 )
 
 print(tokenizer.decode(outputs[0], skip_special_tokens=True))