explosion/spaCy
1
1
Fork 0
mirror of https://github.com/explosion/spaCy.git synced 2025-10-29 23:17:59 +03:00
Code Issues Packages Projects Releases Wiki Activity

add custom code support to CLI speed benchmark

Browse Source
This commit is contained in:
Eliana Vornov 2024-01-18 12:41:18 -05:00
parent e2a3952de5
commit 5f32a5a82e
2 changed files with 20 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
spacy/cli/benchmark_speed.py
View File

@ -13,7 +13,7 @@ from .. import util
from ..language import Language
from ..tokens import Doc
from ..training import Corpus
from ._util import Arg, Opt, benchmark_cli, setup_gpu
from ._util import Arg, Opt, benchmark_cli, setup_gpu, import_code
@benchmark_cli.command(
@ -30,12 +30,14 @@ def benchmark_speed_cli(
use_gpu: int = Opt(-1, "--gpu-id", "-g", help="GPU ID or -1 for CPU"),
n_batches: int = Opt(50, "--batches", help="Minimum number of batches to benchmark", min=30,),
warmup_epochs: int = Opt(3, "--warmup", "-w", min=0, help="Number of iterations over the data for warmup"),
code_path: Optional[Path] = Opt(None, "--code", "-c", help="Path to Python file with additional code (registered functions) to be imported"),
# fmt: on
):
"""
Benchmark a pipeline. Expects a loadable spaCy pipeline and benchmark
data in the binary .spacy format.
"""
import_code(code_path)
setup_gpu(use_gpu=use_gpu, silent=False)
nlp = util.load_model(model)
@ -171,5 +173,7 @@ def print_outliers(sample: numpy.ndarray):
def warmup(
nlp: Language, docs: List[Doc], warmup_epochs: int, batch_size: Optional[int]
) -> numpy.ndarray:
docs = warmup_epochs * docs
return annotate(nlp, docs, batch_size)
warmup_docs = []
for _ in range(warmup_epochs):
warmup_docs += [doc.from_docs([doc]) for doc in docs]
return annotate(nlp, warmup_docs, batch_size)

5
website/docs/api/cli.mdx
View File

@ -1268,13 +1268,14 @@ the [binary `.spacy` format](/api/data-formats#binary-training). The pipeline is
warmed up before any measurements are taken.
```cli
$ python -m spacy benchmark speed [model] [data_path] [--batch_size] [--no-shuffle] [--gpu-id] [--batches] [--warmup]
$ python -m spacy benchmark speed [model] [data_path] [--code] [--batch_size] [--no-shuffle] [--gpu-id] [--batches] [--warmup]
```
| Name | Description |
| -------------------- | -------------------------------------------------------------------------------------------------------- |
| -------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| `model` | Pipeline to benchmark the speed of. Can be a package or a path to a data directory. ~~str (positional)~~ |
| `data_path` | Location of benchmark data in spaCy's [binary format](/api/data-formats#training). ~~Path (positional)~~ |
| `--code`, `-c` | Path to Python file with additional code to be imported. Allows [registering custom functions](/usage/training#custom-functions) for new architectures. ~~Optional[Path] \(option)~~ |
| `--batch-size`, `-b` | Set the batch size. If not set, the pipeline's batch size is used. ~~Optional[int] \(option)~~ |
| `--no-shuffle` | Do not shuffle documents in the benchmark data. ~~bool (flag)~~ |
| `--gpu-id`, `-g` | GPU to use, if any. Defaults to `-1` for CPU. ~~int (option)~~ |