Instructions to use devJy/SEEDRA-zero-small with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use devJy/SEEDRA-zero-small with Transformers:

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

pipe = pipeline("text-generation", model="devJy/SEEDRA-zero-small")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("devJy/SEEDRA-zero-small")
model = AutoModelForCausalLM.from_pretrained("devJy/SEEDRA-zero-small")
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 devJy/SEEDRA-zero-small with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/devJy/SEEDRA-zero-small

SGLang

How to use devJy/SEEDRA-zero-small 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 "devJy/SEEDRA-zero-small" \
    --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": "devJy/SEEDRA-zero-small",
		"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 "devJy/SEEDRA-zero-small" \
        --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": "devJy/SEEDRA-zero-small",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use devJy/SEEDRA-zero-small with Docker Model Runner:
```
docker model run hf.co/devJy/SEEDRA-zero-small
```

devJy commited on Apr 21, 2025

Commit

d2c7f25

verified ·

1 Parent(s): 3c83169

Update README.md

Browse files

Files changed (1) hide show

README.md +3 -3

README.md CHANGED Viewed

@@ -10,7 +10,7 @@ language:
 ---
 # SEEDRA
-![img](./assets/seedra.png)
 ## MODEL DOWNLOAD
@@ -43,13 +43,13 @@ By contrast, smaller “core” models (1 B, 3 B, 8 B, 14 B) often struggle
 ## How to use SEEDRA
-![img](./assets/tta.png)
 1. If you need to build a special instruction dataset or have a domain-specific training dataset, you can augment it using SEEDRA.
 2. For example, Qwen 2.5 is a very powerful model, strong in Chinese and English but somewhat weak in Korean. In such cases, you can use SEEDRA to augment Korean data—using varied expressions and sentence orders to secure more subtoken coverage and strengthen its Korean capabilities.
 3. In a multimodal workflow, you can convert SEEDRA‑generated text into speech using a tool like OuteTTS, then feed that synthesized audio into an ASR (automatic speech recognition) model as additional training or validation data.
-![img](./assets/mma.png)
 ## DEMO

 ---
 # SEEDRA
+![img](./seedra.png)
 ## MODEL DOWNLOAD
 ## How to use SEEDRA
+![img](./tta.png)
 1. If you need to build a special instruction dataset or have a domain-specific training dataset, you can augment it using SEEDRA.
 2. For example, Qwen 2.5 is a very powerful model, strong in Chinese and English but somewhat weak in Korean. In such cases, you can use SEEDRA to augment Korean data—using varied expressions and sentence orders to secure more subtoken coverage and strengthen its Korean capabilities.
 3. In a multimodal workflow, you can convert SEEDRA‑generated text into speech using a tool like OuteTTS, then feed that synthesized audio into an ASR (automatic speech recognition) model as additional training or validation data.
+![img](./mma.png)
 ## DEMO