Instructions to use CPSPX/babylm-zho-pinyin-code-33M with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use CPSPX/babylm-zho-pinyin-code-33M with Transformers:

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

pipe = pipeline("text-generation", model="CPSPX/babylm-zho-pinyin-code-33M", trust_remote_code=True)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("CPSPX/babylm-zho-pinyin-code-33M", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use CPSPX/babylm-zho-pinyin-code-33M with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "CPSPX/babylm-zho-pinyin-code-33M"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "CPSPX/babylm-zho-pinyin-code-33M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/CPSPX/babylm-zho-pinyin-code-33M

SGLang

How to use CPSPX/babylm-zho-pinyin-code-33M 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 "CPSPX/babylm-zho-pinyin-code-33M" \
    --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": "CPSPX/babylm-zho-pinyin-code-33M",
		"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 "CPSPX/babylm-zho-pinyin-code-33M" \
        --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": "CPSPX/babylm-zho-pinyin-code-33M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use CPSPX/babylm-zho-pinyin-code-33M with Docker Model Runner:
```
docker model run hf.co/CPSPX/babylm-zho-pinyin-code-33M
```

CPSPX commited on Jun 12

Commit

81830df

verified ·

1 Parent(s): 577b90a

Add evaluation config.yaml (locked NLU finetune hparams)

Browse files

Files changed (1) hide show

config.yaml +32 -0

config.yaml ADDED Viewed

	@@ -0,0 +1,32 @@

+# Chinese BabyLM 2026 — EVALUATION config for CPSPX/babylm-zho-pinyin-code-33M.
+# This is the eval-pipeline config (NOT a training-params file). It fixes the
+# NLU-track fine-tuning hyperparameters used at evaluation time.
+models:
+  - path: CPSPX/babylm-zho-pinyin-code-33M
+    backend: causal
+tasks:
+  zero_shot:
+    - zhoblimp
+    - hanzi_structure
+    - hanzi_pinyin
+  cogbench:
+    - word_fmri
+    - fmri
+  finetune:
+    - afqmc
+    - ocnli
+    - tnews
+    - cluewsc2020
+eval_dir: evaluation_data
+results_dir: results
+save_item_with_unk: true
+# NLU-track fine-tuning hyperparameters (locked before hidden tasks).
+finetune_hparams:
+  lr: 3.0e-5
+  batch_size: 64
+  max_epochs: 5
+  sequence_length: 256
+  seed: 42