Instructions to use jimfhahn/bibframe-olmo3-v3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use jimfhahn/bibframe-olmo3-v3 with Transformers:

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

pipe = pipeline("text-generation", model="jimfhahn/bibframe-olmo3-v3")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("jimfhahn/bibframe-olmo3-v3")
model = AutoModelForCausalLM.from_pretrained("jimfhahn/bibframe-olmo3-v3", 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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use jimfhahn/bibframe-olmo3-v3 with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf jimfhahn/bibframe-olmo3-v3:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf jimfhahn/bibframe-olmo3-v3:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf jimfhahn/bibframe-olmo3-v3:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf jimfhahn/bibframe-olmo3-v3:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf jimfhahn/bibframe-olmo3-v3:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf jimfhahn/bibframe-olmo3-v3:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf jimfhahn/bibframe-olmo3-v3:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf jimfhahn/bibframe-olmo3-v3:Q4_K_M

Use Docker

docker model run hf.co/jimfhahn/bibframe-olmo3-v3:Q4_K_M

LM Studio
Jan

vLLM

How to use jimfhahn/bibframe-olmo3-v3 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "jimfhahn/bibframe-olmo3-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": "jimfhahn/bibframe-olmo3-v3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/jimfhahn/bibframe-olmo3-v3:Q4_K_M

SGLang

How to use jimfhahn/bibframe-olmo3-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 "jimfhahn/bibframe-olmo3-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": "jimfhahn/bibframe-olmo3-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 "jimfhahn/bibframe-olmo3-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": "jimfhahn/bibframe-olmo3-v3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use jimfhahn/bibframe-olmo3-v3 with Ollama:
```
ollama run hf.co/jimfhahn/bibframe-olmo3-v3:Q4_K_M
```
Unsloth Desktop

How to use jimfhahn/bibframe-olmo3-v3 with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf jimfhahn/bibframe-olmo3-v3:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "jimfhahn/bibframe-olmo3-v3:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use jimfhahn/bibframe-olmo3-v3 with Docker Model Runner:
```
docker model run hf.co/jimfhahn/bibframe-olmo3-v3:Q4_K_M
```

Lemonade

How to use jimfhahn/bibframe-olmo3-v3 with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull jimfhahn/bibframe-olmo3-v3:Q4_K_M

Run and chat with the model

lemonade run user.bibframe-olmo3-v3-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use jimfhahn/bibframe-olmo3-v3 with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf jimfhahn/bibframe-olmo3-v3:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default jimfhahn/bibframe-olmo3-v3:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use jimfhahn/bibframe-olmo3-v3 with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf jimfhahn/bibframe-olmo3-v3:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "jimfhahn/bibframe-olmo3-v3:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

BIBFRAME OLMo-3 7B (v3)

The open model powering BF Vibe, a desktop assistant that helps catalogers create and repair valid BIBFRAME RDF/XML. The model has three roles inside the application wherein BF Vibe generates BIBFRAME structure deterministically and uses SHACL validation as the acceptance test:

Correction — repair a record that fails SHACL validation, grounded in the validator's own report (/fix and the silent post-create refine loop)
Intent routing — classify free-text requests into the app's closed command registry with a calibrated confidence (strict JSON)
Field extraction — pull bibliographic fields out of a description as strict JSON for the app's deterministic record generator

Model details

Developed by: Jim Hahn
Base model: allenai/Olmo-3-7B-Instruct (Apache-2.0; fully open data + weights + recipe lineage)
Method: QLoRA (r=32, α=64), 8K context, effective batch 16, lr 1e-4, 3 epochs; adapter merged into the base weights (this repo ships the merged model)
Training data: jimfhahn/bf-vibe-bibframe-corrections — ≈78% SHACL-verified (corrupted → conforming) BIBFRAME pairs from Library of Congress records, ≈11% intent routing, ≈11% grounded tool traces
License: Apache 2.0

Evaluation

Intent-routing calibration (46-utterance held-out panel, hard-blocklisted from training data generation): accuracy 91.3%, ECE 0.065; at the application's 0.8 confidence gate: proceed-precision 92.3%, false-proceed 0.0% (the model never confidently acts on an out-of-scope request; unconfident requests fall to a clarifying chooser in the app).

Intended use and limitations

Designed for BIBFRAME 2.x RDF/XML following Library of Congress conventions, operating inside a system that (a) generates record structure deterministically and (b) accepts a repair only if SHACL validation strictly improves. Using raw generations without that validation gate is not the intended use.
Repair coverage is not complete (47.5% conforming on held-out hard cases); the surrounding application keeps the original record when no strictly better version is found.
English-centric prompts; source records skew toward LoC cataloging practice.

How to use

Best use case is within the BF Vibe app.

GGUF quantizations are provided with the embedded chat template deliberately stripped: OLMo-3's built-in Jinja template uses tojson, which Ollama's tool-parser cannot compile (the runner exits on load). This repo instead ships Ollama-standard template and params files carrying the validated ChatML + Hermes-style tools template and sampling parameters, so ollama pull hf.co/… picks them up automatically. If you use the GGUF with another llama.cpp runtime, apply the template/params files (or your runtime's ChatML preset) rather than relying on embedded metadata. The safetensors + chat_template.jinja in this repo are unaffected — Jinja consumers (transformers, vLLM) can use them as-is.

Provenance

Source records: id.loc.gov (US public domain)
Validation: the public SHACL service at validate.bibframe.app

Acknowledgements

This work made use of the Illinois Campus Cluster, a computing resource that is operated by the Illinois Campus Cluster Program (ICCP) in conjunction with the National Center for Supercomputing Applications (NCSA) and which is supported by funds from the University of Illinois Urbana-Champaign.

Downloads last month: 248

Safetensors

Model size

7B params

Tensor type

BF16

Model tree for jimfhahn/bibframe-olmo3-v3

Base model

allenai/Olmo-3-1025-7B

Finetuned

allenai/Olmo-3-7B-Instruct-SFT

Finetuned

allenai/Olmo-3-7B-Instruct-DPO

Finetuned

allenai/Olmo-3-7B-Instruct

Quantized

(40)

this model

jimfhahn
/

bibframe-olmo3-v3

BIBFRAME OLMo-3 7B (v3)

Model details

Evaluation

Intended use and limitations

How to use

Provenance

Acknowledgements

Model tree for jimfhahn/bibframe-olmo3-v3

Dataset used to train jimfhahn/bibframe-olmo3-v3