Source: https://seerepo.com/compare/FareedKhan-dev/train-llm-from-scratch/vs/vllm-project/vllm
Markdown source: https://seerepo.com/compare/FareedKhan-dev/train-llm-from-scratch/vs/vllm-project/vllm.md
Description: train-llm-from-scratch and vllm share 5 technologies: torchaudio, torchvision, torch, tiktoken, and datasets. Both are written in Python.

# train-llm-from-scratch vs vllm

train-llm-from-scratch and vllm share 5 technologies: torchaudio, torchvision, torch, tiktoken, and datasets. Both are written in Python.

- [FareedKhan-dev/train-llm-from-scratch](https://seerepo.com/r/FareedKhan-dev/train-llm-from-scratch): A straightforward method for training your LLM, from downloading data to generating text.
- [vllm-project/vllm](https://seerepo.com/r/vllm-project/vllm): A high-throughput and memory-efficient inference and serving engine for LLMs

## What they share and what sets them apart

- In common: torchaudio, torchvision, torch, tiktoken, datasets
- Only in train-llm-from-scratch: h5py
- Only in vllm: FastAPI, axum, gitdb, pyzmq, Hugging Face, Tokio, starlette, Ray, AIOHTTP, tower, mkdocs-material, indicatif, AMD, apache-tvm-ffi, auto-round-lib, conch-triton-kernels, cuda-bindings, cuda-python, cupy-cuda12x, cupy-cuda13x, flashinfer-python, humming-kernels, lm-format-enforcer, mooncake-transfer-engine, mooncake-transfer-engine-rocm, NVIDIA, nvidia-cudnn-frontend, OpenAI, openai-harmony, oss-harmony, peft, quack-kernels, sentencepiece, vllm-bench, vllm-chat, vllm-engine-core-client, vllm-llm, vllm-managed-engine, vllm-parser, vllm-proto, vllm-server, vllm-text, vllm-tokenizer, transformers, xgrammar-structural-tag, einops, llm-multimodal, compressed-tensors, safetensors, triton, mistral-common, xgrammar, partial-json-parser, lmcache, timm, vllm-xpu-kernels, runai-model-streamer, tokenizers, tokenspeed-mla, llguidance, MCP, outlines-core, tpu-inference, fastokens, fastsafetensors, hf-hub, instanttensor, nixl, riptoken, tekken, tensorizer, tiktoken-rs, markdown, markupsafe, mkdocs-material-extensions, parquet, properdocs, pymdown-extensions, selectolax, xlsxwriter, pillow, image, jinja2, minijinja, torchcodec, minijinja-contrib, annotated-types, Pydantic, msgspec, validator, jsonschema, pydantic-core, opentelemetry-api, opentelemetry-semantic-conventions-ai, python-json-logger, vllm-metrics, vllm-tracing, tracing, prometheus-client, psutil, tracing-futures, tracing-subscriber, model-hosting-container-standards, OpenTelemetry, opentelemetry-sdk, prometheus-fastapi-instrumentator

## Side by side

| | train-llm-from-scratch | vllm |
|---|---|---|
| Size | 12,346 lines written, 79 code files | 2,426,215 lines written, 3,378 code files |
| Who writes it | 7.1% of its commits signed by AI. With Claude Code and GitHub Copilot | 13.4% of its commits signed by AI. With Claude Code, Codex, Gemini CLI, Cursor, GitHub Copilot, Devin, Mistral Vibe, CodeRabbit, Factory, and Qwen Code |
| Activity | Last push October 2, 2026, 155 commits, 13,179 stars | Last push November 9, 2026, 24,920 commits, 96,584 stars |
| Tests | 26 test files. 39.2% of its code files have a test that imports them | 2,469 test files. 60.9% 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, train-llm-from-scratch or vllm?

vllm, with 2,426,215 lines written; train-llm-from-scratch has 12,346.

The proof: [vllm-project/vllm's code at commit 821851e](https://github.com/vllm-project/vllm/tree/821851eab038d2ec1bcdcb8c7da32a3728f71358)

### Which has more GitHub stars, train-llm-from-scratch or vllm?

vllm, with 96,584 stars; train-llm-from-scratch has 13,179.

The proof: [vllm-project/vllm on GitHub](https://github.com/vllm-project/vllm)

### Which has more tests, train-llm-from-scratch or vllm?

vllm, with 2,469 test files; train-llm-from-scratch has 26.

The proof: [vllm-project/vllm's code at commit 821851e](https://github.com/vllm-project/vllm/tree/821851eab038d2ec1bcdcb8c7da32a3728f71358)

### Which was updated more recently, train-llm-from-scratch or vllm?

vllm: its last push was on November 9, 2026, and train-llm-from-scratch's on October 2, 2026.

The proof: [vllm-project/vllm's commits on GitHub](https://github.com/vllm-project/vllm/commits)

### Which has more commits signed by AI, train-llm-from-scratch or vllm?

vllm: an AI signed 13.4% of its commits, with Claude Code, Codex, Gemini CLI, Cursor, GitHub Copilot, Devin, Mistral Vibe, CodeRabbit, Factory, and Qwen Code; in train-llm-from-scratch, 7.1%.

The proof: [the commits each tool signed in vllm-project/vllm](https://seerepo.com/r/vllm-project/vllm/ai)

## Sources

- FareedKhan-dev/train-llm-from-scratch: Read on November 10, 2026, at commit bdd480a (https://github.com/FareedKhan-dev/train-llm-from-scratch/tree/bdd480a518094ca834e05ab32131dc126b522d3d)
- vllm-project/vllm: Read on November 9, 2026, at commit 821851e (https://github.com/vllm-project/vllm/tree/821851eab038d2ec1bcdcb8c7da32a3728f71358)

## 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.