Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more.
Python agentaichatgptdeepseekfine-tuninggemmaimage-generationllamallmllmsopenaipythonqwenreinforcement-learningself-hostedstable-diffusiontext-to-speechttsuiunsloth
# SPDX-License-Identifier: AGPL-3.0-only
# Copyright 2026-present the Unsloth AI Inc. team. All rights reserved. See /studio/LICENSE.AGPL-3.0
"""Real GPU generation smoke that skips on GitHub-hosted runners; select it with -k gpu_generation."""
from __future__ import annotations
import pytest
from real_accelerator import (
has_real_cuda,
)
torch = pytest.importorskip("torch")
MODEL_ID = "unsloth/gemma-3-270m-it"
MIN_NEW_TOKENS = 4
MAX_NEW_TOKENS = 16
@pytest.mark.skipif(not has_real_cuda(), reason = "requires a CUDA GPU")
def test_gpu_generation_smoke():
try:
from transformers import AutoModelForCausalLM, AutoTokenizer
except Exception as exc: # pragma: no cover - env without transformers
pytest.skip(f"transformers unavailable: {exc}")
# Gemma emits only <pad> in fp16; use bf16 where supported, else fp32.
dtype = torch.bfloat16 if torch.cuda.is_bf16_supported() else torch.float32
try:
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
model = AutoModelForCausalLM.from_pretrained(MODEL_ID, dtype = dtype).to("cuda")
except Exception as exc:
pytest.skip(f"could not fetch/load {MODEL_ID}: {exc}")
model.eval()
messages = [{"role": "user", "content": "Say hello in one word."}]
inputs = tokenizer.apply_chat_template(
messages, add_generation_prompt = True, return_dict = True, return_tensors = "pt"
).to("cuda")
prompt_len = inputs["input_ids"].shape[1]
with torch.no_grad():
output = model.generate(
**inputs,
min_new_tokens = MIN_NEW_TOKENS,
max_new_tokens = MAX_NEW_TOKENS,
do_sample = False,
)
assert output.shape[1] > prompt_len, "no tokens were generated on the GPU"
reply = tokenizer.decode(output[0][prompt_len:], skip_special_tokens = True)
assert reply.strip(), "expected a non-empty GPU generation"