Instructions to use FredyRivera-dev/MiniCPM5-1B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use FredyRivera-dev/MiniCPM5-1B with Transformers:

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

pipe = pipeline("text-generation", model="FredyRivera-dev/MiniCPM5-1B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("FredyRivera-dev/MiniCPM5-1B")
model = AutoModelForCausalLM.from_pretrained("FredyRivera-dev/MiniCPM5-1B", 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

vLLM

How to use FredyRivera-dev/MiniCPM5-1B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "FredyRivera-dev/MiniCPM5-1B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FredyRivera-dev/MiniCPM5-1B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/FredyRivera-dev/MiniCPM5-1B

SGLang

How to use FredyRivera-dev/MiniCPM5-1B 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 "FredyRivera-dev/MiniCPM5-1B" \
    --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": "FredyRivera-dev/MiniCPM5-1B",
		"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 "FredyRivera-dev/MiniCPM5-1B" \
        --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": "FredyRivera-dev/MiniCPM5-1B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use FredyRivera-dev/MiniCPM5-1B with Docker Model Runner:
```
docker model run hf.co/FredyRivera-dev/MiniCPM5-1B
```

FredyRivera-dev commited on 1 day ago

Commit

29ca31f

verified ·

1 Parent(s): fd91d5b

Update README.md

Browse files

Files changed (1) hide show

README.md +8 -5

README.md CHANGED Viewed

@@ -21,11 +21,14 @@ datasets:
   - openbmb/UltraData-SFT-2605
 ---
-> ## Changes from the original checkpoint
-> Base model: [`openbmb/MiniCPM5-1B`](https://huggingface.co/openbmb/MiniCPM5-1B). Weights are unmodified, no fine-tuning was performed. Only two preparatory changes were made:
-> - **Tokenizer**: the reserved, never-trained slot `<unused_token_0>` (id `130082`) was renamed to `<image>` and marked as a special token.
-> - **Chat template**: now accepts `content` as a list of blocks (`[{"type": "image"}, {"type": "text", ...}]`) and inserts `<image>` at the corresponding position.
-> **This model has no vision capability.** No image encoder or projector is attached. The `<image>` token has no learned meaning (its embedding is ~0) until it's trained alongside an actual vision module. This repo is only a prepared starting point for that future integration.
 <div align="center">
 <img src="https://raw.githubusercontent.com/OpenBMB/MiniCPM/main/assets/minicpm_logo.png" width="500em" />

   - openbmb/UltraData-SFT-2605
 ---
+## Changes from the original checkpoint
+Base model: [`openbmb/MiniCPM5-1B`](https://huggingface.co/openbmb/MiniCPM5-1B). Weights are unmodified, no fine-tuning was performed. Only two preparatory changes were made:
+- **Tokenizer**: the reserved, never-trained slot `<unused_token_0>` (id `130082`) was renamed to `<image>` and marked as a special token.
+- **Chat template**: now accepts `content` as a list of blocks (`[{"type": "image"}, {"type": "text", ...}]`) and inserts `<image>` at the corresponding position.
+**This model has no vision capability.** No image encoder or projector is attached. The `<image>` token has no learned meaning (its embedding is ~0) until it's trained alongside an actual vision module. This repo is only a prepared starting point for that future integration.
 <div align="center">
 <img src="https://raw.githubusercontent.com/OpenBMB/MiniCPM/main/assets/minicpm_logo.png" width="500em" />