Instructions to use SZLHOLDINGS/forge-v3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SZLHOLDINGS/forge-v3 with Transformers:

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

pipe = pipeline("text-generation", model="SZLHOLDINGS/forge-v3")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("SZLHOLDINGS/forge-v3")
model = AutoModelForMultimodalLM.from_pretrained("SZLHOLDINGS/forge-v3", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.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(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use SZLHOLDINGS/forge-v3 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SZLHOLDINGS/forge-v3"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SZLHOLDINGS/forge-v3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/SZLHOLDINGS/forge-v3

SGLang

How to use SZLHOLDINGS/forge-v3 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 "SZLHOLDINGS/forge-v3" \
    --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": "SZLHOLDINGS/forge-v3",
		"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 "SZLHOLDINGS/forge-v3" \
        --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": "SZLHOLDINGS/forge-v3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use SZLHOLDINGS/forge-v3 with Docker Model Runner:
```
docker model run hf.co/SZLHOLDINGS/forge-v3
```

forge-v3 / README.md

betterwithage

restore: revert off-route 2026-11-14 card stamp to audited revision 1ab75c516c4c

ecbb3b3 verified about 7 hours ago

preview code

Raw History Blame Contribute Delete

2.68 kB

---
pipeline_tag: text-generation
license: apache-2.0
library_name: transformers
tags:
- szl-holdings
- doctrine-v11
- launch-contract-governed
- research-candidate
base_model: SZLHOLDINGS/szl-receiptagent-qwen35-0.8b-sft-v3
base_model_revision: 0340c300052e248367a3407aa493ef72a0da79fe
---
<div align="center">

<img src="https://huggingface.co/SZLHOLDINGS/SZLHOLDINGS/resolve/main/assets/mark.svg" width="110" alt="SZL"/>

# forge-v3

**Doctrine-corpus-v2 SFT student (estate doctrine, source-grounded)**

`TRAINED THROUGH THE FAIL-CLOSED LAUNCH CONTRACT` · `RESEARCH_CANDIDATE` · `NOT_PROMOTED`

</div>

---

## Proof of training

![launch-contract](https://huggingface.co/SZLHOLDINGS/SZLHOLDINGS/resolve/main/assets/launch-contract.svg)

## Lineage — every pin immutable

| Stage | Artifact | Pin |
|---|---|---|
| Student base | `szl-receiptagent-qwen35-0.8b-sft-v3` | `0340c300` |
| Corpus | Doctrine-corpus-v2 SFT student (estate doctrine, source-grounded) | group-disjoint split, seed 3407 |
| Admission | operator receipt, `ADMITTED_FOR_TRAINING`, Apache 2.0 | on file |
| Trainer | szl-forge governed pipeline | [launch contract #708](https://github.com/szl-holdings/szl-forge/pull/708) |
| Receipts | [model-eval-receipts](https://huggingface.co/datasets/SZLHOLDINGS/model-eval-receipts) | per-wave |

## Honest status

- ✅ Trained through the launch contract: immutable pins, disjoint holdout, corpus floor, admission receipt
- ❌ Not promoted · not quality-certified · no broad benchmark
- 📊 Evaluation on the group-disjoint holdout: pending publication

<div align="center"><sub>SZL Holdings · Doctrine v11 LOCKED · failures stay published</sub></div>


## The estate

| Surface | Link |
|---|---|
| Company | [szlholdings.com](https://szlholdings.com) |
| Sponsors | [Become a sponsor](https://szlholdings.com/sponsors.html) |
| GitHub | [szl-holdings](https://github.com/szl-holdings) |
| Receipts chain | [frontier-54-runs](https://huggingface.co/datasets/SZLHOLDINGS/frontier-54-runs) |
| Live dashboard | [SI command center](https://szlholdings-szlholdings-si.hf.space) |

## Card completeness (2026-11-12)

This section adds required card sections the card above did not have. It states only what files in this repository and the estate's own receipt datasets show; nothing here is a new result.

### Evaluation

Not yet evaluated. As of 2026-11-12 this repository has no evaluation results, battery receipts or `model-index` entries, so this card makes no quality, accuracy or safety claim.

The estate model registry (`frontier-54-runs/registry/model-registry.json`, generated 2026-11-10) classifies this repo as `HOLD (weights, no provenance)`.

	---
	pipeline_tag: text-generation
	license: apache-2.0
	library_name: transformers
	tags:
	- szl-holdings
	- doctrine-v11
	- launch-contract-governed
	- research-candidate
	base_model: SZLHOLDINGS/szl-receiptagent-qwen35-0.8b-sft-v3
	base_model_revision: 0340c300052e248367a3407aa493ef72a0da79fe
	---

	<div align="center">

	<img src="https://huggingface.co/SZLHOLDINGS/SZLHOLDINGS/resolve/main/assets/mark.svg" width="110" alt="SZL"/>

	# forge-v3

	Doctrine-corpus-v2 SFT student (estate doctrine, source-grounded)

	`TRAINED THROUGH THE FAIL-CLOSED LAUNCH CONTRACT` · `RESEARCH_CANDIDATE` · `NOT_PROMOTED`

	</div>

	---

	## Proof of training

	![launch-contract](https://huggingface.co/SZLHOLDINGS/SZLHOLDINGS/resolve/main/assets/launch-contract.svg)

	## Lineage — every pin immutable

	\| Stage \| Artifact \| Pin \|
	\|---\|---\|---\|
	\| Student base \| `szl-receiptagent-qwen35-0.8b-sft-v3` \| `0340c300` \|
	\| Corpus \| Doctrine-corpus-v2 SFT student (estate doctrine, source-grounded) \| group-disjoint split, seed 3407 \|
	\| Admission \| operator receipt, `ADMITTED_FOR_TRAINING`, Apache 2.0 \| on file \|
	\| Trainer \| szl-forge governed pipeline \| [launch contract #708](https://github.com/szl-holdings/szl-forge/pull/708) \|
	\| Receipts \| [model-eval-receipts](https://huggingface.co/datasets/SZLHOLDINGS/model-eval-receipts) \| per-wave \|

	## Honest status

	- ✅ Trained through the launch contract: immutable pins, disjoint holdout, corpus floor, admission receipt
	- ❌ Not promoted · not quality-certified · no broad benchmark
	- 📊 Evaluation on the group-disjoint holdout: pending publication

	<div align="center"><sub>SZL Holdings · Doctrine v11 LOCKED · failures stay published</sub></div>


	## The estate

	\| Surface \| Link \|
	\|---\|---\|
	\| Company \| [szlholdings.com](https://szlholdings.com) \|
	\| Sponsors \| [Become a sponsor](https://szlholdings.com/sponsors.html) \|
	\| GitHub \| [szl-holdings](https://github.com/szl-holdings) \|
	\| Receipts chain \| [frontier-54-runs](https://huggingface.co/datasets/SZLHOLDINGS/frontier-54-runs) \|
	\| Live dashboard \| [SI command center](https://szlholdings-szlholdings-si.hf.space) \|

	## Card completeness (2026-11-12)

	This section adds required card sections the card above did not have. It states only what files in this repository and the estate's own receipt datasets show; nothing here is a new result.

	### Evaluation

	Not yet evaluated. As of 2026-11-12 this repository has no evaluation results, battery receipts or `model-index` entries, so this card makes no quality, accuracy or safety claim.

	The estate model registry (`frontier-54-runs/registry/model-registry.json`, generated 2026-11-10) classifies this repo as `HOLD (weights, no provenance)`.