Spaces:
Running
Running
File size: 9,132 Bytes
b247dc4 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 |
"""Test the HuggingFace API."""
import math
import os
from subprocess import PIPE, Popen
import numpy as np
import pytest
from manifest.api.models.huggingface import MODEL_REGISTRY, TextGenerationModel
from manifest.api.models.sentence_transformer import SentenceTransformerModel
NOCUDA = 0
try:
p = Popen(
[
"nvidia-smi",
(
"--query-gpu=index,utilization.gpu,memory.total,memory.used,"
"memory.free,driver_version,name,gpu_serial,display_active,"
"display_mode"
),
"--format=csv,noheader,nounits",
],
stdout=PIPE,
)
except OSError:
NOCUDA = 1
MAXGPU = 0
if NOCUDA == 0:
try:
p = os.popen( # type: ignore
"nvidia-smi --query-gpu=index --format=csv,noheader,nounits"
)
i = p.read().split("\n") # type: ignore
MAXGPU = int(i[-2]) + 1
except OSError:
NOCUDA = 1
def test_load_non_registry_model() -> None:
"""Test load model not in registry."""
model_name = "NinedayWang/PolyCoder-160M"
assert model_name not in MODEL_REGISTRY
model = TextGenerationModel(
model_name_or_path=model_name, model_type="text-generation"
)
result = model.generate("Why is the sky green?", max_tokens=10)
assert result is not None
def test_gpt_generate() -> None:
"""Test pipeline generation from a gpt model."""
model = TextGenerationModel(
model_name_or_path="gpt2",
use_accelerate=False,
use_parallelize=False,
use_bitsandbytes=False,
use_deepspeed=False,
use_fp16=False,
device=-1,
)
inputs = "Why is the sky green?"
result = model.generate(inputs, max_tokens=10)
assert result is not None
assert len(result) == 1
assert result[0][0] == "\n\nThe sky is green.\n\nThe"
assert math.isclose(round(result[0][1], 3), -11.516)
result = model.generate("Cats are", max_tokens=10)
assert result is not None
assert len(result) == 1
assert result[0][0] == " not the only ones who are being targeted by the"
assert math.isclose(round(result[0][1], 3), -21.069)
result = model.generate(inputs, max_tokens=5)
assert result is not None
assert len(result) == 1
assert result[0][0] == "\n\nThe sky is"
assert math.isclose(round(result[0][1], 3), -6.046)
# Truncate max length
model.pipeline.max_length = 5
result = model.generate(inputs, max_tokens=2)
assert result is not None
assert len(result) == 1
assert result[0][0] == "\n\n"
assert math.isclose(round(result[0][1], 3), -1.414)
def test_encdec_generate() -> None:
"""Test pipeline generation from a gpt model."""
model = TextGenerationModel(
model_name_or_path="google/t5-small-lm-adapt",
use_accelerate=False,
use_parallelize=False,
use_bitsandbytes=False,
use_deepspeed=False,
use_fp16=False,
device=-1,
)
inputs = "Why is the sky green?"
result = model.generate(inputs, max_tokens=10)
assert result is not None
assert len(result) == 1
assert result[0][0] == "What is the sky green? What is the sky"
assert math.isclose(round(result[0][1], 3), -7.271)
result = model.generate("Cats are", max_tokens=10)
assert result is not None
assert len(result) == 1
assert result[0][0] == "a great way to get out of the house"
assert math.isclose(round(result[0][1], 3), -13.868)
result = model.generate(inputs, max_tokens=5)
assert result is not None
assert len(result) == 1
assert result[0][0] == "What is the sky green"
assert math.isclose(round(result[0][1], 3), -5.144)
# Truncate max length
model.pipeline.max_length = 5
result = model.generate(inputs, max_tokens=2)
assert result is not None
assert len(result) == 1
assert result[0][0] == "Is"
assert math.isclose(round(result[0][1], 3), -4.233)
def test_gpt_score() -> None:
"""Test pipeline generation from a gpt model."""
model = TextGenerationModel(
model_name_or_path="gpt2",
use_accelerate=False,
use_parallelize=False,
use_bitsandbytes=False,
use_deepspeed=False,
use_fp16=False,
device=-1,
)
inputs = ["Why is the sky green?", "Cats are butterflies"]
result = model.score_sequence(inputs)
assert result is not None
assert len(result) == 2
assert math.isclose(round(result[0][0], 3), -46.71)
assert math.isclose(round(result[1][0], 3), -12.752)
assert isinstance(result[0][1], list)
assert isinstance(result[1][1], list)
def test_embed() -> None:
"""Test embedding pipeline."""
model = TextGenerationModel(
model_name_or_path="gpt2",
use_accelerate=False,
use_parallelize=False,
use_bitsandbytes=False,
use_deepspeed=False,
use_fp16=False,
device=-1,
)
inputs = ["Why is the sky green?", "Cats are butterflies"]
embeddings = model.embed(inputs)
assert isinstance(embeddings, np.ndarray)
assert embeddings.shape == (2, 768)
model2 = SentenceTransformerModel(
model_name_or_path="all-mpnet-base-v2",
use_accelerate=False,
use_parallelize=False,
use_bitsandbytes=False,
use_deepspeed=False,
use_fp16=False,
device=-1,
)
inputs = ["Why is the sky green?", "Cats are butterflies"]
embeddings = model2.embed(inputs)
assert isinstance(embeddings, np.ndarray)
assert embeddings.shape == (2, 768)
def test_batch_gpt_generate() -> None:
"""Test pipeline generation from a gpt model."""
model = TextGenerationModel(
model_name_or_path="gpt2",
use_accelerate=False,
use_parallelize=False,
use_bitsandbytes=False,
use_deepspeed=False,
use_fp16=False,
device=-1,
)
inputs = ["Why is the sky green?", "Cats are"]
result = model.generate(inputs, max_tokens=10)
assert result is not None
assert len(result) == 2
assert result[0][0] == "\n\nThe sky is green.\n\nThe"
assert math.isclose(round(result[0][1], 3), -11.516)
assert result[1][0] == " not the only ones who are being targeted by the"
assert math.isclose(round(result[1][1], 3), -21.069)
result = model.generate(inputs, max_tokens=5)
assert result is not None
assert len(result) == 2
assert result[0][0] == "\n\nThe sky is"
assert math.isclose(round(result[0][1], 2), -6.05)
assert result[1][0] == " not the only ones who"
assert math.isclose(round(result[1][1], 3), -9.978)
# Truncate max length
model.pipeline.max_length = 5
result = model.generate(inputs, max_tokens=2)
assert result is not None
assert len(result) == 2
assert result[0][0] == "\n\n"
assert math.isclose(round(result[0][1], 3), -1.414)
assert result[1][0] == " not the"
assert math.isclose(round(result[1][1], 3), -6.246)
def test_batch_encdec_generate() -> None:
"""Test pipeline generation from a gpt model."""
model = TextGenerationModel(
model_name_or_path="google/t5-small-lm-adapt",
use_accelerate=False,
use_parallelize=False,
use_bitsandbytes=False,
use_deepspeed=False,
use_fp16=False,
device=-1,
)
inputs = ["Why is the sky green?", "Cats are"]
result = model.generate(inputs, max_tokens=10)
assert result is not None
assert len(result) == 2
assert result[0][0] == "What is the sky green? What is the sky"
assert math.isclose(round(result[0][1], 3), -7.271)
assert result[1][0] == "a great way to get out of the house"
assert math.isclose(round(result[1][1], 3), -13.868)
result = model.generate(inputs, max_tokens=5)
assert result is not None
assert len(result) == 2
assert result[0][0] == "What is the sky green"
assert math.isclose(round(result[0][1], 3), -5.144)
assert result[1][0] == "a great way to"
assert math.isclose(round(result[1][1], 3), -6.353)
# Truncate max length
model.pipeline.max_length = 5
result = model.generate(inputs, max_tokens=2)
assert result is not None
assert len(result) == 2
assert result[0][0] == "Is"
assert math.isclose(round(result[0][1], 3), -4.233)
assert result[1][0] == "a"
assert math.isclose(round(result[1][1], 3), -1.840)
@pytest.mark.skipif(
(NOCUDA == 1 or MAXGPU == 0), reason="No cuda or GPUs found through nvidia-smi"
)
def test_gpt_deepspeed_generate() -> None:
"""Test deepspeed generation from a gpt model."""
model = TextGenerationModel(
model_name_or_path="gpt2",
use_accelerate=False,
use_parallelize=False,
use_bitsandbytes=False,
use_deepspeed=True,
use_fp16=False,
device=0,
)
inputs = "Why is the sky green?"
result = model.generate(inputs, max_tokens=10)
assert result is not None
assert len(result) == 1
assert result[0][0] == "\n\nThe sky is green.\n\nThe"
assert math.isclose(round(result[0][1], 3), -11.517)
|