Instructions to use blackdaqlabs/MiloX1-60M with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use blackdaqlabs/MiloX1-60M with Transformers:

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

pipe = pipeline("text-generation", model="blackdaqlabs/MiloX1-60M")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("blackdaqlabs/MiloX1-60M")
model = AutoModelForCausalLM.from_pretrained("blackdaqlabs/MiloX1-60M", 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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use blackdaqlabs/MiloX1-60M with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "blackdaqlabs/MiloX1-60M"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "blackdaqlabs/MiloX1-60M",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/blackdaqlabs/MiloX1-60M

SGLang

How to use blackdaqlabs/MiloX1-60M 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 "blackdaqlabs/MiloX1-60M" \
    --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": "blackdaqlabs/MiloX1-60M",
		"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 "blackdaqlabs/MiloX1-60M" \
        --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": "blackdaqlabs/MiloX1-60M",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use blackdaqlabs/MiloX1-60M with Docker Model Runner:
```
docker model run hf.co/blackdaqlabs/MiloX1-60M
```

MiloX

MiloX es un modelo de lenguaje tipo GPT (Transformer decoder, arquitectura Llama: RMSNorm + RoPE + SwiGLU + GQA) de 62M parámetros, preentrenado desde cero (sin pesos de ningún modelo previo) y afinado para conversar. Su asistente se llama Milo.

Contexto: 1024 tokens · Vocabulario: 16384 (BPE byte-level propio) · Idiomas: español e inglés
Formato de chat: <|system|>, <|user|>, <|assistant|> terminados en <|end|> (ya incluido como chat_template)

Uso

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM

repo = "TU_USUARIO/MiloX"
device = "cuda" if torch.cuda.is_available() else "cpu"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(repo).to(device).eval()

msgs = [{"role": "user", "content": "Hola, ¿quién eres?"}]
text = tok.apply_chat_template(msgs, tokenize=False, add_generation_prompt=True)
inputs = tok(text, return_tensors="pt", add_special_tokens=False).to(device)
out = model.generate(**inputs, max_new_tokens=200, do_sample=True, temperature=0.5,
                     top_p=0.9, top_k=30)
print(tok.decode(out[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))

Limitaciones

Es un modelo pequeño: puede equivocarse, inventar datos o repetirse. No lo uses para decisiones importantes. Revisa las licencias de los datasets usados en el preentrenamiento y el fine-tuning antes de elegir la licencia final del repo.

Downloads last month: 1,081

Safetensors

Model size

61.9M params

Tensor type

F16

blackdaqlabs
/

MiloX1-60M

MiloX

Uso

Limitaciones

Space using blackdaqlabs/MiloX1-60M 1