TensorRT-LLMs/tests/unittest/_torch/speculative/test_ngram.py
Yan Chunwei 5506f60037
chore [BREAKING CHANGE]: Flatten PyTorchConfig knobs into TorchLlmArgs (#4603)
Signed-off-by: Superjomn <328693+Superjomn@users.noreply.github.com>
2025-05-28 18:43:04 +08:00

83 lines
2.6 KiB
Python

import os
import sys
import unittest
import pytest
import torch
from tensorrt_llm import SamplingParams
from tensorrt_llm._torch import LLM
from tensorrt_llm.llmapi import KvCacheConfig, NGramDecodingConfig
sys.path.append(os.path.join(os.path.dirname(__file__), '..'))
from utils.llm_data import llm_models_root
# TODO: Add cuda graph enabled tests.
# Cuda graph cannot currently be enabled for ngram because cuda graph requires
# spec metadata and ngram does not have it.
@pytest.mark.parametrize("use_cuda_graph,attn_backend",
[[False, "TRTLLM"], [False, "FLASHINFER"]])
def test_llama_ngram(use_cuda_graph: bool, attn_backend: str):
total_mem_gb = torch.cuda.get_device_properties(0).total_memory / 1e9
if total_mem_gb < 31:
pytest.skip("Not enough memory to load target model")
models_path = llm_models_root()
pytorch_config = dict(
enable_overlap_scheduler=False,
use_cuda_graph=use_cuda_graph,
# Only create a single CUDA graph to prevent OOM in CI
attn_backend=attn_backend,
cuda_graph_batch_sizes=[1],
)
kv_cache_config = KvCacheConfig(enable_block_reuse=False, max_tokens=2080)
sampling_params = SamplingParams(
max_tokens=32,
temperature=0,
)
max_batch_size = 1
target_model_dir = f"{models_path}/llama-models-v2/llama-v2-13b-hf"
draft_len = 4
spec_config = NGramDecodingConfig(
prompt_lookup_num_tokens=draft_len,
max_matching_ngram_size=draft_len,
is_keep_all=True,
is_use_oldest=True,
is_public_pool=True,
)
llm_spec = LLM(model=target_model_dir,
max_batch_size=max_batch_size,
**pytorch_config,
kv_cache_config=kv_cache_config,
speculative_config=spec_config)
prompts = [
"The capital of France is", "The president of the United States is"
]
results_spec = llm_spec.generate(prompts, sampling_params)
generated_text_spec = [result.outputs[0].text for result in results_spec]
llm_spec.shutdown()
llm_ref = LLM(model=target_model_dir,
max_batch_size=max_batch_size,
**pytorch_config,
kv_cache_config=kv_cache_config)
results_ref = llm_ref.generate(prompts, sampling_params)
generated_text_ref = [result.outputs[0].text for result in results_ref]
llm_ref.shutdown()
for text_spec, text_ref in zip(generated_text_spec, generated_text_ref):
# The spec decode algorithm currently guarantees identical results
assert text_spec == text_ref
if __name__ == "__main__":
unittest.main()