Instructions to use malaiwah/GLM-5.3-Flash-TR3-8bpw with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use malaiwah/GLM-5.3-Flash-TR3-8bpw with Transformers:

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

pipe = pipeline("image-text-to-text", model="malaiwah/GLM-5.3-Flash-TR3-8bpw")
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, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("malaiwah/GLM-5.3-Flash-TR3-8bpw")
model = AutoModelForMultimodalLM.from_pretrained("malaiwah/GLM-5.3-Flash-TR3-8bpw", 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=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Trellis

How to use malaiwah/GLM-5.3-Flash-TR3-8bpw with Trellis:

# No code snippets available yet for this library.

# To use this model, check the repository files and the library's documentation.

# Want to help? PRs adding snippets are welcome at:
# https://github.com/huggingface/huggingface.js

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use malaiwah/GLM-5.3-Flash-TR3-8bpw with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "malaiwah/GLM-5.3-Flash-TR3-8bpw"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "malaiwah/GLM-5.3-Flash-TR3-8bpw",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/malaiwah/GLM-5.3-Flash-TR3-8bpw

SGLang

How to use malaiwah/GLM-5.3-Flash-TR3-8bpw 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 "malaiwah/GLM-5.3-Flash-TR3-8bpw" \
    --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": "malaiwah/GLM-5.3-Flash-TR3-8bpw",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

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 "malaiwah/GLM-5.3-Flash-TR3-8bpw" \
        --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": "malaiwah/GLM-5.3-Flash-TR3-8bpw",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use malaiwah/GLM-5.3-Flash-TR3-8bpw with Docker Model Runner:
```
docker model run hf.co/malaiwah/GLM-5.3-Flash-TR3-8bpw
```

malaiwah commited on 3 days ago

Commit

b5ca3b3

verified ·

1 Parent(s): e51e904

disclose that materialization-receipt.json names a destroyed filesystem

Browse files

packed_root and output_root in materialization-receipt.json are the
PRODUCER's absolute paths on a rented GPU filesystem that no longer exists.
The receipt is self-sealed and its digest is verified against the PUBLISHED
bytes on every measurement run, so correcting the paths in place would
permanently break every future measurement against this release and would
falsify a record of where the encode ran. The receipt is untouched; this
file states what those fields are and names the reading path that works
(--source tr3 / --source exl3hf).

Files changed (1) hide show

MATERIALIZATION-PATHS.md +45 -0

MATERIALIZATION-PATHS.md ADDED Viewed

	@@ -0,0 +1,45 @@

+# `materialization-receipt.json` names paths that are not on your machine
+**Added:** 2026-08-29, as a correction. Nothing in this repository was edited to add it.
+`materialization-receipt.json` in this repository records
+    "packed_root": "/home/jl_fs/glm53-k8/out-k8"
+    "output_root": "/home/jl_fs/glm53-k8/ckpt-k8"
+Those are the **producer's absolute paths on a rented GPU filesystem that no longer
+exists**. They are sealed provenance metadata describing where the encode ran. They are
+not resolvable locations, and nothing you can do locally will make them resolve.
+## Why they were not simply corrected
+The receipt is self-sealed: `receipt_sha256` is a SHA-256 over its own canonical form
+with that field blanked. Editing any byte changes the digest, and that digest is verified
+against the **published bytes** on every measurement run
+(`k6/tools/tr3_surface.py::verify_seal`, called from `bin/measure_cloud.py`, which
+raises `this release's PUBLISHED seal does not reproduce`). Rewriting the paths to make
+them prettier would permanently break every future measurement against this release, and
+would falsify a record of where the encode actually ran. The honest fix for a wrong
+pointer inside a sealed record is a correction beside it, not a quiet edit.
+## What to read instead
+| You want to | Use |
+|---|---|
+| Score this release | `--source tr3` (or `--source exl3hf`) — reads the published shards directly and never consults `packed_root` |
+| Verify the release's own claims | `k6/tools/tr3_surface.py` — recomputes all 12 claims in the receipt from the published bytes before decoding |
+| Reproduce the published KLD | the published capture dataset, `malaiwah/GLM-5.3-Flash-fidelity-suite-v1` |
+`--source checkpoint` and `--source payload-store` are **producer-side** reading paths.
+The payload store was never published, so they are not reachable from this repository —
+that is a publishing gap, not something a flag can fix. `--source checkpoint` resolves
+`packed_root` out of this receipt and will fail on the path above; that failure is the
+expected outcome, not a misconfiguration on your side.
+## Schema note
+This receipt's `schema` is `malaiwah.glm53-k8-materialization-receipt.v1`.
+Its sibling release uses the other family string for the same document type
+(`quant-pipeline.glm53-k6-...` vs `malaiwah.glm53-k8-...`). Both are historical; no
+validator in the suite keys on this field. Recorded here so a third party writing one
+does not assume the namespace is stable.