glm-5.3-flash-tiny-random-nvfp4

Instructions to use Coder40-95/glm-5.3-flash-tiny-random-nvfp4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Coder40-95/glm-5.3-flash-tiny-random-nvfp4 with Transformers:

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

pipe = pipeline("image-text-to-text", model="Coder40-95/glm-5.3-flash-tiny-random-nvfp4")
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("Coder40-95/glm-5.3-flash-tiny-random-nvfp4")
model = AutoModelForMultimodalLM.from_pretrained("Coder40-95/glm-5.3-flash-tiny-random-nvfp4", 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 Coder40-95/glm-5.3-flash-tiny-random-nvfp4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Coder40-95/glm-5.3-flash-tiny-random-nvfp4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Coder40-95/glm-5.3-flash-tiny-random-nvfp4",
		"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/Coder40-95/glm-5.3-flash-tiny-random-nvfp4

SGLang

How to use Coder40-95/glm-5.3-flash-tiny-random-nvfp4 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 "Coder40-95/glm-5.3-flash-tiny-random-nvfp4" \
    --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": "Coder40-95/glm-5.3-flash-tiny-random-nvfp4",
		"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 "Coder40-95/glm-5.3-flash-tiny-random-nvfp4" \
        --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": "Coder40-95/glm-5.3-flash-tiny-random-nvfp4",
		"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 Coder40-95/glm-5.3-flash-tiny-random-nvfp4 with Docker Model Runner:
```
docker model run hf.co/Coder40-95/glm-5.3-flash-tiny-random-nvfp4
```

glm-5.3-flash-tiny-random-nvfp4

File size: 1,411 Bytes

963e33f

--- flashinfer/mla/_sparse_mla_sm120.py
+++ flashinfer/mla/_sparse_mla_sm120.py
@@ -112,6 +112,7 @@
         (64, 512),
         (64, 1024),
         (64, 2048),
+        (64, 2176),
         (128, 128),
         (128, 512),
         (128, 1024),
--- flashinfer/data/csrc/sparse_mla_sm120_decode_dsv3_2.cu
+++ flashinfer/data/csrc/sparse_mla_sm120_decode_dsv3_2.cu
@@ -162,6 +164,7 @@
   DSV3_2_DISPATCH(64, 512)
   DSV3_2_DISPATCH(64, 1024)
   DSV3_2_DISPATCH(64, 2048)
+  DSV3_2_DISPATCH(64, 2176)
   DSV3_2_DISPATCH(128, 128)
   DSV3_2_DISPATCH(128, 512)
   DSV3_2_DISPATCH(128, 1024)
--- flashinfer/data/csrc/sparse_mla_sm120_prefill.cu
+++ flashinfer/data/csrc/sparse_mla_sm120_prefill.cu
@@ -217,6 +217,16 @@
                          float* out_lse, float sm_scale, int num_tokens, size_t stride_kv_block,
                          const int* topk_length_ptr, cudaStream_t stream) {
   static_assert(KVCacheTraits<MT>::D_QK == 576);
+  if (topk == 2176) {
+    if constexpr (MT == ModelType::GLM_NSA) {
+      if (num_heads != 64) return false;
+      launch_prefill_mg<MT, ComputeMode::FP8, 64, 2176, 64>(
+          Q, KV, indices, attn_sink, output, out_lse, sm_scale, num_tokens,
+          stride_kv_block, topk_length_ptr, stream);
+      return true;
+    }
+    return false;
+  }
   if (topk != 2048) return false;
 
   // PBS=64 matches the V32 decode (`decode_dsv3_2_kernel.cuh`). NH=8 covers