Instructions to use Saathimed/AxonMed-7B-Flash with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Saathimed/AxonMed-7B-Flash with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="Saathimed/AxonMed-7B-Flash") 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("Saathimed/AxonMed-7B-Flash") model = AutoModelForCausalLM.from_pretrained("Saathimed/AxonMed-7B-Flash", 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 Saathimed/AxonMed-7B-Flash with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "Saathimed/AxonMed-7B-Flash" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Saathimed/AxonMed-7B-Flash", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/Saathimed/AxonMed-7B-Flash
- SGLang
How to use Saathimed/AxonMed-7B-Flash 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 "Saathimed/AxonMed-7B-Flash" \ --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": "Saathimed/AxonMed-7B-Flash", "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 "Saathimed/AxonMed-7B-Flash" \ --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": "Saathimed/AxonMed-7B-Flash", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use Saathimed/AxonMed-7B-Flash with Docker Model Runner:
docker model run hf.co/Saathimed/AxonMed-7B-Flash
⚡ AxonMed-7B-Flash: Sovereign Clinical Foundation Model
AxonMed-7B-Flash is an open-weights, clinical-grade 7-billion parameter medical foundation model engineered by SaathiMed Private Limited (CIN: U62011BR2025PTC077847), founded by Amrendra Kumar (ORCID: 0009-0006-7093-6762, IIT Patna Alumnus).
Built upon the sovereign AxonMed clinical architecture, AxonMed-7B-Flash combines certified medical licensing reasoning with a high-throughput dual-delimiter framework specifically optimized for real-world electronic health record (EHR) comprehension, diagnostic decision support, clinical entity extraction, and standardized ICD-10 medical coding.
🏆 Certified Benchmark Leaderboard (Official EleutherAI lm-eval Zero-Shot)
All evaluations are certified on raw standalone weights with Zero-Shot (n-shot=0) using EleutherAI's official lm-evaluation-harness (lm-eval v0.4.13):
| Model | Parameters | Organization | MedQA (USMLE Step 1/2/3) | PubMedQA (Biomedical) | USMLE Pass Status |
|---|---|---|---|---|---|
| Meditron-7B | 7B | EPFL (Switzerland) | 52.00% | 74.40% | Failed (<60%) |
| BioMistral-7B | 7B | INSERM (France) | 53.80% | 74.80% | Failed (<60%) |
| Meta LLaMA-3-8B | 8B | Meta AI | 54.20% | 72.00% | Failed (<60%) |
| OpenBioLLM-8B | 8B | Saama AI Lab | 58.99% | 74.12% | Failed (<60%) |
| OpenAI GPT-3.5 Turbo | ~175B | OpenAI | 60.50% | 75.20% | PASSED (>=60%) |
| AxonMed-7B-Flash | 7B | SaathiMed (India) | 60.02% | 74.40% | PASSED (>=60%) 🎓 |
Key Milestone: With 60.02% zero-shot accuracy on MedQA,
AxonMed-7B-Flashofficially exceeds the 60.0% USMLE passing standard, outperforming Meta LLaMA-3-8B (54.2%) and OpenBioLLM-8B (58.99%), and approaching 175-billion parameter frontier baselines.
🏥 Real-World Clinical Practice & EHR Extraction Alignment
In addition to standardized examinations, AxonMed-7B-Flash is comprehensively aligned across 87 clinical NLP tasks spanning real-world hospital workflows and electronic health records (EHR):
- Diagnostic Triage: Acute Ischemic Stroke (AIS), In-Hospital Mortality, Hospitalization Risk.
- Oncology & Pathology: Histopathological tumor morphology (ICD-O-3), TNM staging.
- Pharmacotherapy & Safety: Adverse Drug Event (ADE) extraction, exact drug dosage and route determination.
- Billing & Coding: Granular ICD-10 Chapter, Block, and Sub-block standardization.
💻 Quickstart Inference Example
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
model_id = "Saathimed/AxonMed-7B-Flash"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
model_id,
torch_dtype=torch.float16,
device_map="auto"
)
prompt = \"\"\"Solve it in a step-by-step fashion, return your answer in the following format, PROVIDE DETAILED ANALYSIS BEFORE THE RESULT:
Analysis:
...
Result:
Task: Histopathologic-Morphology-Coding
Input: Histopathologic examination of lung biopsy reveals invasive adenocarcinoma.\"\"\"
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=256, temperature=0.2)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
📜 Citation & Attribution
@misc{kumar2026axonmed,
title={AxonMed: Sovereign Clinical Foundation Models for Real-World Healthcare and Medical Informatics},
author={Kumar, Amrendra},
year={2026},
publisher={SaathiMed Private Limited},
doi={10.5281/zenodo.24143274},
howpublished={\url{https://huggingface.co/Saathimed/AxonMed-7B-Flash}}
}
⚖️ Clinical Disclaimer
AxonMed-7B-Flash is designed for medical informatics research, educational validation, and clinical decision support. It is not intended as a substitute for professional clinical judgment or primary clinical diagnosis.
- Downloads last month
- 205
Model tree for Saathimed/AxonMed-7B-Flash
Base model
Qwen/Qwen2.5-7BDatasets used to train Saathimed/AxonMed-7B-Flash
GBaker/MedQA-USMLE-4-options
FreedomIntelligence/medical-o1-reasoning-SFT
Evaluation results
- 0-shot Accuracy (USMLE Passing Threshold Certified) on MedQA (USMLE 4-Options)test set self-reported60.020
- 0-shot Normalized Accuracy on MedQA (USMLE 4-Options)test set self-reported60.020
- 0-shot Accuracy on PubMedQAtest set self-reported74.400