Instructions to use baidu/ERNIE-5.5-Flash-Base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use baidu/ERNIE-5.5-Flash-Base with Transformers:

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

pipe = pipeline("text-generation", model="baidu/ERNIE-5.5-Flash-Base")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("baidu/ERNIE-5.5-Flash-Base", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use baidu/ERNIE-5.5-Flash-Base with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "baidu/ERNIE-5.5-Flash-Base"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "baidu/ERNIE-5.5-Flash-Base",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/baidu/ERNIE-5.5-Flash-Base

SGLang

How to use baidu/ERNIE-5.5-Flash-Base 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 "baidu/ERNIE-5.5-Flash-Base" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "baidu/ERNIE-5.5-Flash-Base",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "baidu/ERNIE-5.5-Flash-Base" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "baidu/ERNIE-5.5-Flash-Base",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use baidu/ERNIE-5.5-Flash-Base with Docker Model Runner:
```
docker model run hf.co/baidu/ERNIE-5.5-Flash-Base
```

Interesting, its benchmark scores completely crush Deepseek

by SuperLogic - opened 9 days ago

Discussion

SuperLogic

9 days ago

Interesting, its benchmark scores completely crush Deepseek, and the same goes for Hy4. The soon-to-be-released Qwen4 as well, putting the pressure back on Deepseek again.

xmesaj2

8 days ago

it's deepsek v4, not v4.1 and not 1107

sjy1203

BAIDU org 6 days ago

Interesting, its benchmark scores completely crush Deepseek, and the same goes for Hy4. The soon-to-be-released Qwen4 as well, putting the pressure back on Deepseek again.

Thanks! The chart compares similar-size bases: DeepSeek-V4-Flash-Base, Hy3-preview-Base and MiMo-V2.5-Base. We lead on 24 of 30, not all. Our attention builds on DeepSeek's MLA, DSA and HCA, so credit to them too. Where did you see Qwen4 base scores? We haven't found any yet.

sjy1203

BAIDU org 6 days ago

it's deepsek v4, not v4.1 and not 1107

Right, we used V4-Flash-Base, closest in size. 1017 is a post-training update, so DeepSeek's latest base is V4.1-Flash-Base. By their numbers we're higher on MMLU-Pro, MGSM, BBH and LongBench-v2 despite their 552B + 196B Engram vs our 280B.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment