Instructions to use randhir302/HumanFlow with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use randhir302/HumanFlow with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="randhir302/HumanFlow")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("randhir302/HumanFlow")
model = AutoModelForCausalLM.from_pretrained("randhir302/HumanFlow", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use randhir302/HumanFlow with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "randhir302/HumanFlow"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "randhir302/HumanFlow",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/randhir302/HumanFlow

SGLang

How to use randhir302/HumanFlow with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "randhir302/HumanFlow" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "randhir302/HumanFlow",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "randhir302/HumanFlow" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "randhir302/HumanFlow",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use randhir302/HumanFlow with Docker Model Runner:
```
docker model run hf.co/randhir302/HumanFlow
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

HumanFlow-Llama3-8B

Humanize AI Text with Natural Structure, Flow & Tone

🤗 Hugging Face Model • 💻 GitHub Repository • Apache-2.0 License

Overview

HumanFlow-Llama3-8B is a fine-tuned Llama 3 model designed to transform robotic AI-generated writing into content that feels natural, human, readable, and authentic.

Instead of replacing words only, HumanFlow improves:

sentence rhythm
structure
tone
flow
readability
realism

Why HumanFlow?

Most AI-generated text feels:

repetitive
over-polished
generic
predictable
emotionally flat

HumanFlow rewrites outputs to feel more organic and naturally written.

Performance Snapshot

Metric	Base Model	HumanFlow
Human-Like Score	18%	99%
Natural Tone	Low	High
Rewrite Quality	Basic	Advanced
Readability	Generic	Strong

Internal Evaluation

Metric	Score
BERTScore F1	0.8424
ROUGE-L	0.0908
Perplexity	1.5242
Text Overlap	0.0528

Best Use Cases

SEO rewriting
Blog enhancement
Student writing cleanup
Email personalization
AI content polishing
SaaS integrations
Human-style generation pipelines

Before vs After

Input

In today’s rapidly evolving digital landscape, it is imperative for organizations to leverage strategic methodologies in order to maximize engagement.

HumanFlow Output

Online markets move fast. If a company wants attention, it needs smart strategy, clear messaging, and content people actually care about.

Quickstart

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

model_id = "randhir302/HumanFlow"

tokenizer = AutoTokenizer.from_pretrained(model_id)

model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.float16,
    device_map="auto"
)

prompt = """
Rewrite this in a more human tone:

Artificial intelligence is transforming industries worldwide.
"""

inputs = tokenizer(prompt, return_tensors="pt").to("cuda")

outputs = model.generate(
    **inputs,
    max_new_tokens=220,
    temperature=0.75,
    top_p=0.90,
    repetition_penalty=1.10
)

print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Recommended Settings

temperature = 0.75
top_p = 0.90
repetition_penalty = 1.10
max_new_tokens = 700

## Roadmap

- [x] Public Launch  
- [x] Hugging Face Release  
- [x] Fine-Tuned Base Model  
- [ ] GGUF Quantized Release  
- [ ] HumanFlow Pro API  
- [ ] Browser Editor  
- [ ] Multilingual Version  

---

## Community

If HumanFlow helps you:

⭐ Like the model  
⭐ Share outputs  
⭐ Benchmark it  
⭐ Build products with it

Downloads last month: 64

Safetensors

Model size

8B params

Tensor type

BF16

Model tree for randhir302/HumanFlow

Base model

unsloth/llama-3-8b-Instruct-bnb-4bit

Finetuned

(1,274)

this model

Quantizations

6 models

Evaluation results

BERTScore F1 on Internal Evaluation Suite
self-reported

0.842
ROUGE-L on Internal Evaluation Suite
self-reported

0.091
Perplexity on Internal Evaluation Suite
self-reported

1.524
Text Overlap on Internal Evaluation Suite
self-reported

0.053