Instructions to use Abu-Sameer-66/SciPeerAI-7B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Abu-Sameer-66/SciPeerAI-7B with Transformers:

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

pipe = pipeline("text-generation", model="Abu-Sameer-66/SciPeerAI-7B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("Abu-Sameer-66/SciPeerAI-7B", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Abu-Sameer-66/SciPeerAI-7B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Abu-Sameer-66/SciPeerAI-7B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Abu-Sameer-66/SciPeerAI-7B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Abu-Sameer-66/SciPeerAI-7B

SGLang

How to use Abu-Sameer-66/SciPeerAI-7B 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 "Abu-Sameer-66/SciPeerAI-7B" \
    --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": "Abu-Sameer-66/SciPeerAI-7B",
		"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 "Abu-Sameer-66/SciPeerAI-7B" \
        --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": "Abu-Sameer-66/SciPeerAI-7B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Abu-Sameer-66/SciPeerAI-7B with Docker Model Runner:
```
docker model run hf.co/Abu-Sameer-66/SciPeerAI-7B
```

SciPeerAI-7B

World's first LLM fine-tuned specifically for scientific fraud detection.

Built by Sameer Nadeem — BS Data Science student, Bahawalpur, Pakistan.

What This Model Does

SciPeerAI-7B analyzes scientific papers and outputs a structured JSON report covering 14 fraud dimensions simultaneously — something no other model does.

Given a paper's title, authors, abstract, and text, it returns:

Overall fraud confidence score (0.0 to 1.0)
Risk score across 14 modules
Fraud type classification (20 fraud taxonomy types)
Final verdict: FRAUD DETECTED or CLEAN PAPER

Model Details

Field	Value
Base model	mistralai/Mistral-7B-Instruct-v0.3
Fine-tuning method	QLoRA (4-bit, nf4)
LoRA rank	r=16, alpha=32
Target modules	q_proj, k_proj, v_proj, o_proj
Training steps	500
Final training loss	0.2352
Training hardware	Kaggle T4 x2 (32GB RAM)
Training time	147.7 minutes
Dataset	SciPeerBench v1.1 (644 papers)

Performance

Test Paper	Expected	Model Output
Wakefield 1998 (Lancet) — retracted vaccine-autism fraud	FRAUD	fraud_confidence=0.99 ✅
NDM-1 resistance paper (Lancet ID, 2010) — landmark clean paper	CLEAN	fraud_confidence=0.05 ✅

Quick Start

from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import PeftModel
import torch

base = "mistralai/Mistral-7B-Instruct-v0.3"
adapter = "Abu-Sameer-66/SciPeerAI-7B"

tokenizer = AutoTokenizer.from_pretrained(base)
model = AutoModelForCausalLM.from_pretrained(base, torch_dtype=torch.float16, device_map="auto")
model = PeftModel.from_pretrained(model, adapter)

prompt = """<s>[INST] Analyze this scientific paper for fraud and integrity issues:

Title: Your Paper Title
Authors: Author Names
Year: 2024
Journal: Journal Name
Abstract: Your abstract text here...

Provide a detailed JSON analysis with fraud scores across all 14 dimensions. [/INST]
"""

inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
output = model.generate(**inputs, max_new_tokens=300, do_sample=False)
print(tokenizer.decode(output[0], skip_special_tokens=True))

14 Fraud Detection Dimensions

Module	What It Detects
Statistical Audit	p-hacking, sample size issues, round numbers
Figure Forensics	pHash + ELA + brightness manipulation
Methodology Checker	causation claims, missing control groups
Citation Analyzer	self-citation rings, unsupported claims
Reproducibility Scanner	code/data/ethics/preregistration
Novelty Scorer	structural signals + Semantic Scholar API
GRIM Test	mathematically impossible means
SPRITE Test	impossible distributions + SD verification
Granularity Analyzer	digit preference, Benford Law
P-Curve Analyzer	publication bias, p-value clustering
Effect Size Validator	Cohen d, power analysis, inflated effects
Retraction Checker	retracted citations via CrossRef live API
Citation Cartel Detector	citation rings, network manipulation
LLM Paper Detector	burstiness, TTR, uniformity patterns

Part of SciPeerAI System

🌐 Live Web UI: scipeerai-ui.vercel.app
⚡ Live API: web-production-f526d.up.railway.app
📊 Benchmark Dataset: Abu-Sameer-66/SciPeerBench
💻 GitHub Backend: github.com/Abu-Sameer-66/SciPeerAI

Citation

@misc{nadeem2026scipeerai,
  title={SciPeerAI: Multi-dimensional Automated Scientific Integrity Analysis System},
  author={Sameer Nadeem},
  year={2026},
  url={https://huggingface.co/Abu-Sameer-66/SciPeerAI-7B}
}

Downloads last month: -; Downloads are not tracked for this model. How to track

Model tree for Abu-Sameer-66/SciPeerAI-7B

Base model

mistralai/Mistral-7B-v0.3

Finetuned

mistralai/Mistral-7B-Instruct-v0.3