Instructions to use dhanesh-hf/Jarvis-Titan-M3-UltraLong-Adapter with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use dhanesh-hf/Jarvis-Titan-M3-UltraLong-Adapter with Transformers:

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

pipe = pipeline("text-generation", model="dhanesh-hf/Jarvis-Titan-M3-UltraLong-Adapter")

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("dhanesh-hf/Jarvis-Titan-M3-UltraLong-Adapter", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use dhanesh-hf/Jarvis-Titan-M3-UltraLong-Adapter with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "dhanesh-hf/Jarvis-Titan-M3-UltraLong-Adapter"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "dhanesh-hf/Jarvis-Titan-M3-UltraLong-Adapter",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/dhanesh-hf/Jarvis-Titan-M3-UltraLong-Adapter

SGLang

How to use dhanesh-hf/Jarvis-Titan-M3-UltraLong-Adapter 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 "dhanesh-hf/Jarvis-Titan-M3-UltraLong-Adapter" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "dhanesh-hf/Jarvis-Titan-M3-UltraLong-Adapter",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "dhanesh-hf/Jarvis-Titan-M3-UltraLong-Adapter" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "dhanesh-hf/Jarvis-Titan-M3-UltraLong-Adapter",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use dhanesh-hf/Jarvis-Titan-M3-UltraLong-Adapter with Docker Model Runner:
```
docker model run hf.co/dhanesh-hf/Jarvis-Titan-M3-UltraLong-Adapter
```

Jarvis-Titan-M3-UltraLong-Adapter / LICENSE

dhanesh-hf

Add J.A.R.V.I.S. Titan Proprietary Research License (JTRL-v1.0)

b85ac6c verified 17 days ago

Raw History Blame Contribute Delete

4.37 kB

	# J.A.R.V.I.S. TITAN PROPRIETARY RESEARCH LICENSE (JTRL-v1.0)
	Copyright (c) 2026 Dhanesh. All Rights Reserved.
	Autonomous J.A.R.V.I.S. Titan Research & Engineering Program

	---

	### 1. PREAMBLE & STATEMENT OF PROPRIETARY RESEARCH
	This repository, including but not limited to: model weights, tensor shards, adapter matrices, architecture definitions, configuration manifests, curated training datasets, and telemetry registries (collectively, the "Work"), embodies original and proprietary scientific research authored and engineered exclusively by Dhanesh ("Licensor").

	The Work implements novel architectural contributions including:
	* The M1–M4 Tri-Brid Memory Subsystem: Sliding Window Attention (SWA), Salient Exact KV Reservoir buffers, and test-time associative Neural Memory ($M_t$).
	* MAG-3 Adaptive Gating Mechanics: Input-dependent three-way dynamic routing ($g_{\text{local}}, g_{\text{res}}, g_{\text{mem}}$).
	* DeepSeekMoE Upcycled Top-K Architecture: Specialized routed expert representations, isolated shared experts, and load-balanced routing weights.
	* Curated Frontier Reasoning Curricula: Token-packed verified STEM, Olympiad, and executable code training sets.

	---

	### 2. GRANT OF NON-EXCLUSIVE RESEARCH PERMIT
	Subject to the terms and limitations of this License, Licensor grants you a non-exclusive, non-transferable, revocable permit to access, inspect, download, and execute the Work STRICTLY FOR NON-COMMERCIAL ACADEMIC AUDIT, INDEPENDENT SCIENTIFIC EVALUATION, AND BENCHMARK REPLICATION.

	---

	### 3. STRICT PROHIBITIONS & RESERVATION OF RIGHTS

	#### A. Commercial Exploitation Prohibited
	You shall NOT use the Work, in whole or in part, for any commercial purpose, enterprise deployment, monetization, paid inference API, hosted model service, software-as-a-service (SaaS), or revenue-generating activity without a separate, explicit commercial agreement signed by the Licensor.

	#### B. Redistribution & Re-Hosting Prohibited
	You shall NOT redistribute, re-host, mirror, re-upload, sublicense, rent, lease, or resell the model weights, adapters, or dataset shards to any third party, public platform, torrent, cloud bucket, or alternative model hub under any identity.

	#### C. Prohibition Against Synthetic Distillation & Model Inversion
	You shall NOT use outputs, hidden states, activations, or synthetic completions generated by this Work to train, pretrain, fine-tune, distill, or align any competing commercial artificial intelligence model, foundational LLM, or derivative system.

	#### D. Derivative Works & Architectural Theft
	You shall NOT reverse-engineer, extract, or claim authorship over the proprietary gating mechanisms, reservoir filtering logic, or calibrated neural memory state equations. Any derivative checkpoints, fine-tunes, or quantized formats (GGUF, AWQ, EXL2) created for academic evaluation remain bound by this License and must retain full attribution to Dhanesh.

	---

	### 4. ATTRIBUTION & CITATION REQUIREMENT
	Any academic paper, technical report, benchmark evaluation, or public demonstration referencing, evaluating, or utilizing this Work must prominently include the following citation:

	```bibtex
	@misc{jarvis_titan_2026,
	author = {Dhanesh},
	title = {J.A.R.V.I.S. Titan 14.8B MoE: Unified Tri-Brid Neural Memory and Autonomous Reasoning Architecture},
	year = {2026},
	publisher = {Hugging Face},
	howpublished = {\url{https://huggingface.co/dhanesh-hf}}
	}
	```

	---

	### 5. MODEL LINEAGE & THIRD-PARTY RIGHTS
	This Work was developed through novel post-training, upcycling, architectural grafting, and calibration research. Where initialized from open research weights (such as the Qwen2.5 base architecture), those components remain subject to their respective upstream research licenses. All adapted expert parameters, neural memory layers, calibrated routing matrices, and curated datasets constitute original proprietary intellectual property of the Licensor.

	---

	### 6. DISCLAIMER OF WARRANTY & LIMITATION OF LIABILITY
	THE WORK IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND NON-INFRINGEMENT. IN NO EVENT SHALL THE AUTHOR OR COPYRIGHT HOLDER BE LIABLE FOR ANY CLAIM, DAMAGES, OR OTHER LIABILITY ARISING FROM THE USE OF THE WORK.