Instructions to use PYTHAI/granite-4.2-8b-fork with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use PYTHAI/granite-4.2-8b-fork with Transformers:

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

pipe = pipeline("text-generation", model="PYTHAI/granite-4.2-8b-fork")
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("PYTHAI/granite-4.2-8b-fork")
model = AutoModelForCausalLM.from_pretrained("PYTHAI/granite-4.2-8b-fork", 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 PYTHAI/granite-4.2-8b-fork with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "PYTHAI/granite-4.2-8b-fork"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "PYTHAI/granite-4.2-8b-fork",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/PYTHAI/granite-4.2-8b-fork

SGLang

How to use PYTHAI/granite-4.2-8b-fork 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 "PYTHAI/granite-4.2-8b-fork" \
    --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": "PYTHAI/granite-4.2-8b-fork",
		"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 "PYTHAI/granite-4.2-8b-fork" \
        --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": "PYTHAI/granite-4.2-8b-fork",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use PYTHAI/granite-4.2-8b-fork with Docker Model Runner:
```
docker model run hf.co/PYTHAI/granite-4.2-8b-fork
```

granite-4.2-8b-fork / FORK.json

Gregory-L

add the Apache-2.0 text the upstream declares but does not ship

59a48ff verified 28 days ago

Raw History Blame Contribute Delete

2.9 kB

	{
	"kind": "licence-locked pointer fork",
	"source_repo": "ibm-granite/granite-4.2-8b",
	"source_url": "https://huggingface.co/ibm-granite/granite-4.2-8b",
	"source_commit": "f8de16cdcdbc6c779ca517604e050d82cc119e44",
	"forked_at_utc": "2026-09-13T23:55:35Z",
	"licence_tag": "apache-2.0",
	"licence_file": {
	"path": "LICENSE",
	"bytes": 11544,
	"sha256": "bbedc3fda3305820b977265f01b8619d87570a6739de3a5582c3464840f1e57a",
	"provenance": "upstream ships no LICENSE file at this commit; its model card declares license: apache-2.0. The Apache License 2.0 text was added here, byte-identical to the LICENSE in Qwen/Qwen3.8-27B at 1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0, because Apache-2.0 section 4(a) requires a copy of the licence with any redistribution."
	},
	"weights_copied": false,
	"weights_not_copied": {
	"files": 4,
	"bytes": 17583228032
	},
	"parameters": 8791592960,
	"load_weights_with": "from_pretrained('ibm-granite/granite-4.2-8b', revision='f8de16cdcdbc6c779ca517604e050d82cc119e44')",
	"files": [
	{
	"path": "README.md",
	"bytes": 38624,
	"sha256": "01396b1a8596586986fea22cfaa86d0255d19113d71fe09e0b69cc2bfa4307da"
	},
	{
	"path": "chat_template.jinja",
	"bytes": 9013,
	"sha256": "f0ba43f79b3cabca5e5a7584c77aec92f49feae45cdf7779c87d9fc54cd90258"
	},
	{
	"path": "config.json",
	"bytes": 907,
	"sha256": "85611f4e34633d4e148e6a5f64bc2d3a23ebbc2b014bc87fbda764eecf261cd9"
	},
	{
	"path": "generation_config.json",
	"bytes": 207,
	"sha256": "1b9baf0a7959d31d7fb58ff6b506c27d09a73d395cadccc7e6dd1d4b7e43fd9d"
	},
	{
	"path": "granite_thinking_parser.py",
	"bytes": 4649,
	"sha256": "cd5eef99f559f33dc402e0475174a97dcb43724d7e810149c702e9bbc473048a"
	},
	{
	"path": "merges.txt",
	"bytes": 916646,
	"sha256": "b6fe424e334903f7fb84d3a106d9730455f4744b9fe3c21ee136d97a00e72502"
	},
	{
	"path": "model.safetensors.index.json",
	"bytes": 31357,
	"sha256": "d37ecbd9468d7925034dc40e9633509ef746f54d52ac6050e2ffc3e365e31504"
	},
	{
	"path": "model.sig",
	"bytes": 11528,
	"sha256": "0a542061b1d4e5bce9314059115b1d88ee22c5c6637fe024f9d394764949bfc6"
	},
	{
	"path": "special_tokens_map.json",
	"bytes": 563,
	"sha256": "e9435fefd6d838fd9fcbbc44b97a8e3ff322be7f6dfb7e4fd2468586574bb52b"
	},
	{
	"path": "tokenizer.json",
	"bytes": 7153414,
	"sha256": "883975314d58743726347caaf2c32ac89afd5ad57222cffbc4f052f2ab27b2d7"
	},
	{
	"path": "tokenizer_config.json",
	"bytes": 17677,
	"sha256": "2007f26df166ed10a2d6fa881c1c8f3cdd151e451469f59b274897f14449835f"
	},
	{
	"path": "vocab.json",
	"bytes": 1612694,
	"sha256": "a1be0c1e0a89b5bacda6af6f8b63df09668cb8d1f656229ee9123140c9507810"
	}
	],
	"why": "preserve the licence terms of this exact commit and pin the revision; the weights stay at the source and are addressed by commit sha",
	"forked_by": "mindX (PYTHAI) via the node token"
	}