Source: https://seerepo.com/compare/facebookresearch/sam-3d-objects/vs/microsoft/unilm
Markdown source: https://seerepo.com/compare/facebookresearch/sam-3d-objects/vs/microsoft/unilm.md
Description: sam-3d-objects and unilm share 14 technologies: h5py, webdataset, Flask, Gradio, auto-gptq, bitsandbytes, peft, pycocotools, torchaudio, flash-attn, xformers, opencv-python, timm, and jsonlines. Both are written in Python.

# sam-3d-objects vs unilm

sam-3d-objects and unilm share 14 technologies: h5py, webdataset, Flask, Gradio, auto-gptq, bitsandbytes, peft, pycocotools, torchaudio, flash-attn, xformers, opencv-python, timm, and jsonlines. Both are written in Python.

- [facebookresearch/sam-3d-objects](https://seerepo.com/r/facebookresearch/sam-3d-objects): SAM 3D Objects
- [microsoft/unilm](https://seerepo.com/r/microsoft/unilm): Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities

## What they share and what sets them apart

- In common: h5py, webdataset, Flask, Gradio, auto-gptq, bitsandbytes, peft, pycocotools, torchaudio, flash-attn, xformers, opencv-python, timm, jsonlines
- Only in sam-3d-objects: fastavro, hdfs, pymongo, MongoDB, werkzeug, Lightning, pydot, Open3D, seaborn, crcmod, cuda-python, einops-exts, fvcore, librosa, NVIDIA, optimum, scikit-image, sentence-transformers, smplx, gsplat, av, decord, openexr, fqdn, loguru
- Only in unilm: FastAPI, blobfile, tinydb, pyarrow, OpenAI, apex, beir, fschat, pyserini, pytorch-fid, rouge-score, scikit-learn, taming-transformers-rom1504, textdistance, torch, transformers, torchvision, datasets, fairscale, sacrebleu, detectron2, accelerate, sentencepiece, diffusers, deepspeed, torchscale, tokenizers, einops, nltk, triton, pytorch-lightning, sacremoses, tiktoken, vLLM, sklearn, mteb, torchmetrics, evaluate, fastwer, ir-datasets, pycocoevalcap, pytrec-eval, pillow, bs4, lxml, boto3

## Side by side

| | sam-3d-objects | unilm |
|---|---|---|
| Size | 21,153 lines written, 109 code files | 769,203 lines written, 2,556 code files |
| Who writes it | 0% of its commits signed by AI | 0% of its commits signed by AI |
| Activity | Last push June 2, 2026, 14 commits, 7,696 stars | Last push September 21, 2026, 1,092 commits, 22,250 stars |
| Tests | 0 test files | 163 test files. 5.1% of its code files have a test that imports them |

- Size: Lines written by people and agents, without generated ones; code files, without tests.
- Who writes it: A commit counts as AI's when it carries its signature: a Co-Authored-By line, its author name or the merge of its branch.
- Activity: From GitHub and its history, on the day SeeRepo read it.
- Tests: The files that test the code, and the share of code files a test imports.

## Frequently asked questions

### Which is bigger, sam-3d-objects or unilm?

unilm, with 769,203 lines written; sam-3d-objects has 21,153.

The proof: [microsoft/unilm's code at commit 31c5b90](https://github.com/microsoft/unilm/tree/31c5b904ca1bf2afb4c234a6675c683a4e5fc7cd)

### Which has more GitHub stars, sam-3d-objects or unilm?

unilm, with 22,250 stars; sam-3d-objects has 7,696.

The proof: [microsoft/unilm on GitHub](https://github.com/microsoft/unilm)

### Which has more tests, sam-3d-objects or unilm?

unilm, with 163 test files.

The proof: [microsoft/unilm's code at commit 31c5b90](https://github.com/microsoft/unilm/tree/31c5b904ca1bf2afb4c234a6675c683a4e5fc7cd)

### Which was updated more recently, sam-3d-objects or unilm?

unilm: its last push was on September 21, 2026, and sam-3d-objects's on June 2, 2026.

The proof: [microsoft/unilm's commits on GitHub](https://github.com/microsoft/unilm/commits)

## Sources

- facebookresearch/sam-3d-objects: Read on November 19, 2026, at commit f91db41 (https://github.com/facebookresearch/sam-3d-objects/tree/f91db411c50efee93d8db7aeb323885650f6f722)
- microsoft/unilm: Read on November 22, 2026, at commit 31c5b90 (https://github.com/microsoft/unilm/tree/31c5b904ca1bf2afb4c234a6675c683a4e5fc7cd)

## What is your repo built with?

Any public repo, with no account: https://seerepo.com/r/owner/repo/built, with the owner and name it has on GitHub.
