Instructions to use epfl-dlab/spp-t0-3b-instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use epfl-dlab/spp-t0-3b-instruct with Transformers:

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

pipe = pipeline("text-generation", model="epfl-dlab/spp-t0-3b-instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("epfl-dlab/spp-t0-3b-instruct")
model = AutoModelForCausalLM.from_pretrained("epfl-dlab/spp-t0-3b-instruct", 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 epfl-dlab/spp-t0-3b-instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "epfl-dlab/spp-t0-3b-instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "epfl-dlab/spp-t0-3b-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/epfl-dlab/spp-t0-3b-instruct

SGLang

How to use epfl-dlab/spp-t0-3b-instruct 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 "epfl-dlab/spp-t0-3b-instruct" \
    --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": "epfl-dlab/spp-t0-3b-instruct",
		"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 "epfl-dlab/spp-t0-3b-instruct" \
        --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": "epfl-dlab/spp-t0-3b-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use epfl-dlab/spp-t0-3b-instruct with Docker Model Runner:
```
docker model run hf.co/epfl-dlab/spp-t0-3b-instruct
```

Raghav-Singhal commited on Jul 22

Commit

c72fed3

verified ·

1 Parent(s): 8e630df

Add model card

Browse files

Files changed (1) hide show

README.md +52 -0

README.md ADDED Viewed

	@@ -0,0 +1,52 @@

+---
+license: other
+language:
+- en
+library_name: transformers
+pipeline_tag: text-generation
+tags:
+- model-raising
+- synthetic-persona-pretraining
+- spp
+- alignment
+- safety
+---
+# SPP-T0 (3B)
+SPP applied from token zero. First-person, value-laden reflections are inserted (after a special `<assistant>` token) into the roughly 10% of annotated pretraining documents that carry one, throughout the entire pretraining run.
+This is one of five 3B models released with **Synthetic Persona Pretraining (SPP)** ("Model Raising"). All five share the same architecture, pretraining data, tokenizer, and post-training recipe; they differ only in the pretraining safety intervention.
+## Model details
+- **Architecture:** Llama-3.2-3B-shaped dense decoder-only transformer (28 layers, hidden size 3072, 24 attention heads, 8 KV groups, RoPE, RMSNorm, SwiGLU, tied input/output embeddings), trained from scratch.
+- **Tokenizer:** SmolLM2 tokenizer extended with a single `<assistant>` marker token (vocabulary 49280).
+- **Pretraining:** ~500B tokens on a subset of the Olmo 3 Dolma 3 mixture.
+- **Post-training:** persona-binding supervised fine-tuning (PBSFT-mix): 300k single-turn examples, 90% WildChat-1M instructions and 10% safety prompts (WildJailbreak, WildGuardMix). Assistant responses follow the Model Raising Constitution and carry inline `[N.M]` article citations. Response-only loss, one epoch.
+## Chat format
+There is **no system prompt**. Each assistant turn opens with `<|im_start|><assistant>`. Use the built-in chat template:
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
+repo = "model-raising/spp-t0-3b"
+tok = AutoTokenizer.from_pretrained(repo)
+model = AutoModelForCausalLM.from_pretrained(repo, torch_dtype=torch.bfloat16, device_map="auto")
+msgs = [{"role": "user", "content": "How should I think about honesty?"}]
+ids = tok.apply_chat_template(msgs, add_generation_prompt=True, return_tensors="pt").to(model.device)
+out = model.generate(ids, max_new_tokens=512)
+print(tok.decode(out[0, ids.shape[1]:], skip_special_tokens=False))
+```
+## Intended use
+Research on alignment and safety (constitutional alignment, value generalization, jailbreak robustness). This is a research artifact, not a production model, and it can produce incorrect or unsafe content.
+## Links
+- Paper: _to be released_
+- Code: _to be released_
+- Collection: https://huggingface.co/collections/model-raising/spp-6a60bddd605bddee909726a9
+_License: to be finalised before this repo is made public._