Instructions to use RMDWLLC/kaiju-coder-7 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use RMDWLLC/kaiju-coder-7 with Transformers:

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

pipe = pipeline("text-generation", model="RMDWLLC/kaiju-coder-7")
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)

# Load model directly
from transformers import AutoProcessor, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("RMDWLLC/kaiju-coder-7")
model = AutoModelForImageTextToText.from_pretrained("RMDWLLC/kaiju-coder-7")
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=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use RMDWLLC/kaiju-coder-7 with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/RMDWLLC/kaiju-coder-7

SGLang

How to use RMDWLLC/kaiju-coder-7 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 "RMDWLLC/kaiju-coder-7" \
    --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": "RMDWLLC/kaiju-coder-7",
		"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 "RMDWLLC/kaiju-coder-7" \
        --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": "RMDWLLC/kaiju-coder-7",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use RMDWLLC/kaiju-coder-7 with Docker Model Runner:
```
docker model run hf.co/RMDWLLC/kaiju-coder-7
```

kaiju-coder-7 / SOURCE_INVENTORY.md

restokes92

Add files using upload-large-folder tool

9afd28d verified 6 days ago

preview code

raw

history blame contribute delete

5.59 kB

	# Kaiju Source Inventory

	Generated from GitHub source-of-truth repositories plus the requested local RMDW wiki snapshot. This inventory defines what may become Kaiju training data, what is eval-only, and what must stay excluded.

	## Global Training Rules

	- Do not train on raw secrets, API keys, OAuth tokens, cookies, private keys, or credential files.
	- Do not train on closed-model responses from OpenAI, Anthropic, Gemini, or similar providers unless the terms clearly allow it.
	- Do not train on client-specific private data without explicit review and consent.
	- Preserve repository name, commit SHA, source path, license, and reviewer status for every promoted dataset row.

	## GitHub Repository Inventory

	\| Repo \| SHA \| Role \| Training use \| Required gates \| Exclusions \| Notes \|
	\|---\|---\|---\|---\|---\|---\|---\|
	\| [RichardEchols/kaiju-coder](https://github.com/RichardEchols/kaiju-coder) \| `3d57eae92ad5` \| model lab, harness, evals, training scripts \| candidate-after-review \| secret-scan, closed-model-output-check, license-review \| runs, models, .secrets, private datasets, raw logs \| Use repo-owned harnesses, evals, docs, scripts, and curated datasets. Exclude weights, generated runs, and local secrets. \|
	\| [RichardEchols/Kiyomi-7.7.7](https://github.com/RichardEchols/Kiyomi-7.7.7) \| `294b31008135` \| business-owner AI-company module contracts \| candidate-after-review \| secret-scan, closed-model-output-check, private-data-review \| credentials, tokens, private client state, closed-model transcripts \| Use module contracts, templates, acceptance gates, and owner-facing task structure as high-signal business-owner curriculum. \|
	\| [RichardEchols/kiyomi-agent](https://github.com/RichardEchols/kiyomi-agent) \| `b192c910f3f7` \| business OS wrapper and local-agent patterns \| candidate-after-review \| secret-scan, closed-model-output-check, private-data-review \| credentials, tokens, local runtime state, private support logs \| Use architecture, docs, scripts, and safe wrapper patterns. Do not train on runtime secrets or private logs. \|
	\| [RichardEchols/rmdw-site](https://github.com/RichardEchols/rmdw-site) \| `df089dc3b2d3` \| public RMDW offer, site, and conversion surface \| candidate-after-review \| secret-scan, closed-model-output-check, public-copy-review \| environment files, deployment secrets, analytics tokens \| Use public offer copy, app structure, pricing/CTA patterns, and website implementation patterns. \|
	\| [RichardEchols/makotoair](https://github.com/RichardEchols/makotoair) \| `7568f07fea6e` \| client website implementation pattern \| eval-and-patterns-only \| secret-scan, client-data-review, consent-review \| client-specific, contact data, contracts, private business details \| Use as eval/pattern inspiration for local service business sites. Do not bulk-train on client-specific text without explicit review. \|
	\| [RichardEchols/Mezzal-Construction](https://github.com/RichardEchols/Mezzal-Construction) \| `e8f2eede0405` \| client website implementation pattern \| eval-and-patterns-only \| secret-scan, client-data-review, consent-review \| client-specific, contact data, contracts, private business details \| Use as eval/pattern inspiration for premium contractor site work. Do not bulk-train on client-specific text without explicit review. \|
	\| [RichardEchols/rmdw-agent-wiki](https://github.com/RichardEchols/rmdw-agent-wiki) \| `ae1b8e85d3fe` \| RMDW/Kiyomi operational wiki \| selective-reference-only \| secret-scan, credentials-redaction, private-data-review, closed-model-output-check \| credentials.md, customers.md, raw, contracts, private client notes, support logs \| Use only redacted strategy/product notes and documented decisions. Never use raw credentials or private client data. \|

	## Local Source Inventory

	Local files are context snapshots, not the source of truth. Promote local wiki material into training only after explicit review, redaction, and either sync/diff against the GitHub wiki or a documented reviewer exception.

	\| Source \| Path \| Git repo \| Files \| Training use \| Required gates \| Excluded paths present \| Safe reference candidates \| Notes \|
	\|---\|---\|---:\|---:\|---\|---\|---\|---\|---\|
	\| RMDW-Wiki-local \| `/Users/richardecholsai7/Documents/RMDW-Wiki` \| no \| 93 \| selective-reference-only \| secret-scan, credentials-redaction, private-data-review, sync-or-diff-against-github \| credentials.md, customers.md, customers/, raw/ \| README.md, kaiju-coder-build-log.md, kaiju-coder-business-plan.md, kaiju-coder-soul.md, kiyomi-agent-build-log.md, pricing-history.md, product/kiyomi-private-ai-workstation.md, ops/product-ops-automation.md, client-acquisition-engine/README.md \| Use as a local context snapshot only after explicit row-level review. Do not treat unsynced local files as the authoritative training source. \|

	## Training Eligibility Meaning

	- `candidate-after-review`: source can produce training or eval examples only after secret scanning, closed-model-output review, and row-level provenance.
	- `eval-and-patterns-only`: use for hard eval prompts, harness behavior, screenshots, or generalized patterns. Do not bulk-train on client-specific source text.
	- `selective-reference-only`: use narrowly after redaction. Treat credentials, customer notes, and raw operational data as excluded by default.
	- Local snapshots require review against the GitHub source of truth before promotion into dataset rows.

	## Next Dataset Step

	Generate candidate examples only from reviewed paths, attach this inventory SHA or local snapshot data to each row, then run `scripts/validate_training_data.py` before any training run.