ColossalAI/tests/test_infer/test_llama_infer.py

90 lines
3.0 KiB
Python

import os
import pytest
import torch
from packaging import version
import colossalai
from colossalai.inference.tensor_parallel.engine import TPInferEngine
from colossalai.logging import disable_existing_loggers
from colossalai.shardformer import ShardConfig
from colossalai.testing import clear_cache_before_run, parameterize, rerun_if_address_is_in_use, spawn
from tests.kit.model_zoo import model_zoo
os.environ["TRANSFORMERS_NO_ADVISORY_WARNINGS"] = "true"
TPSIZE = 2
BATCH_SIZE = 8
MAX_INPUT_LEN = 12
MAX_OUTPUT_LEN = 100
CUDA_SUPPORT = version.parse(torch.version.cuda) > version.parse("11.5")
def init_to_get_rotary(self, base=10000):
self.config.head_dim_ = self.config.hidden_size // self.config.num_attention_heads
if not hasattr(self.config, "rope_scaling"):
rope_scaling_factor = 1.0
else:
rope_scaling_factor = self.config.rope_scaling.factor if self.config.rope_scaling is not None else 1.0
if hasattr(self.config, "max_sequence_length"):
max_seq_len = self.config.max_sequence_length
elif hasattr(self.config, "max_position_embeddings"):
max_seq_len = self.config.max_position_embeddings * rope_scaling_factor
else:
max_seq_len = 2048 * rope_scaling_factor
base = float(base)
inv_freq = 1.0 / (
base ** (torch.arange(0, self.config.head_dim_, 2, device="cpu", dtype=torch.float32) / self.config.head_dim_)
)
t = torch.arange(max_seq_len + 1024 * 64, device="cpu", dtype=torch.float32) / rope_scaling_factor
freqs = torch.outer(t, inv_freq)
self._cos_cached = torch.cos(freqs).to(torch.float16).cuda()
self._sin_cached = torch.sin(freqs).to(torch.float16).cuda()
return
@parameterize(
"test_config",
[
{
"tp_size": TPSIZE,
}
],
)
def run_llama_test(test_config):
sub_model_zoo = model_zoo.get_sub_registry("transformers_llama_for_casual_lm")
for name, (model_fn, data_gen_fn, _, _, _) in sub_model_zoo.items():
orig_model = model_fn()
init_to_get_rotary(orig_model.model, base=10000)
orig_model = orig_model.half()
data = data_gen_fn()
shard_config = ShardConfig(
enable_tensor_parallelism=True if test_config["tp_size"] > 1 else False, inference_only=True
)
infer_engine = TPInferEngine(orig_model, shard_config, BATCH_SIZE, MAX_INPUT_LEN, MAX_OUTPUT_LEN)
generate_kwargs = dict(do_sample=False)
outputs = infer_engine.generate(data, **generate_kwargs)
assert outputs is not None
def check_llama(rank, world_size, port):
disable_existing_loggers()
colossalai.launch(config={}, rank=rank, world_size=world_size, host="localhost", port=port, backend="nccl")
run_llama_test()
@pytest.mark.skipif(not CUDA_SUPPORT, reason="kv-cache manager engine requires cuda version to be higher than 11.5")
@pytest.mark.dist
@rerun_if_address_is_in_use()
@clear_cache_before_run()
def test_llama():
spawn(check_llama, TPSIZE)
if __name__ == "__main__":
test_llama()