ocr-inference
Guides pix2tex CLI and Python OCR inference from equation images to LaTeX, including images, checkpoints, CPU/CUDA choices, and prediction troubleshooting.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- ocr-inference
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Guides pix2tex CLI and Python OCR inference from equation images to LaTeX, including images, checkpoints, CPU/CUDA choices, and prediction troubleshooting.
- disable-model-invocation
- true
02 资源文件 6
- 参考文档references/api-reference.md2.1 KB
- 参考文档references/cli-reference.md1.3 KB
- 参考文档references/image-preprocessing.md1.4 KB
- 参考文档references/troubleshooting.md1.4 KB
- 脚本scripts/check_pix2tex_cli.py2.0 KB
- 脚本scripts/inspect_pix2tex_image.py2.5 KB
03 SKILL.md 正文
OCR Inference
Use this sub-skill when the user wants LaTeX code from an equation image, wants
to call pix2tex or LatexOCR, or needs to debug inference quality, model
checkpoints, image preprocessing, or CLI flags.
Quick Route
- For the Python class and helper signatures, read references/api-reference.md.
- For CLI commands and flags, read references/cli-reference.md.
- For image size/mode/preprocessing decisions, read references/image-preprocessing.md.
- For common failures, read references/troubleshooting.md.
- Before loading model weights, run scripts/check_pix2tex_cli.py and, for a local image, scripts/inspect_pix2tex_image.py.
Standard CLI Workflow
pix2tex --no-cuda path/to/equation.png
Use --no-cuda for CPU-only environments or deterministic smoke tests. Omit it
only after verifying CUDA PyTorch in the target environment. Add --no-resize
when you do not want the auxiliary image-resizer model to adjust image width.
Use -t/--temperature to control sampling; lower values make repeated
predictions less variable.
Standard Python Workflow
from PIL import Image
from pix2tex.cli import LatexOCR
model = LatexOCR() # may download checkpoints if missing
latex = model(Image.open("equation.png"))
print(latex)
Instantiate LatexOCR() only when checkpoint downloads and compute are allowed.
For CPU-only use, pass an argparse-like object or CLI route with no_cuda=True.
Boundaries
- For GUI screenshot workflows or API serving, use ../interactive-apps-and-api/SKILL.md.
- For dataset creation, formula rendering, or scraping, use ../data-preparation/SKILL.md.
- For training/evaluation or model architecture details, use ../training-and-evaluation/SKILL.md.
04 同仓库更多 Skills
passnet-triton-opt
Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.
passnet-solve
END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.
passnet-skill
PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.
passnet-pattern-fusion
Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.
passnet-orchestrate
Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.
passnet-feedback
Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).