Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
A project to improve skills of large language models
| Date | Stars |
|---|---|
| 2026-07-31 | 1014 |
| 2026-08-03 | 1018 |
| 2026-08-06 | 1018 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# Nemo Skills
> [!NOTE]
> **Benchmarks and rollouts are moving to [NeMo-Gym](https://github.com/NVIDIA-NeMo/Gym).** We
> encourage you to move your rollout workflows over to NeMo-Gym, and to implement new benchmarks
> there rather than in Nemo-Skills.
>
> Nemo-Skills will continue to add support for running NeMo-Gym benchmarks, so you can run them with
> the same cluster configs, server configurations, and slurm setup you already use. The benchmark
> implementation itself should live in NeMo-Gym. An early version of this exists today as
> `ns nemo_gym_rollouts`, but the interface is experimental and will likely change.
Nemo-Skills is a collection of pipelines to improve "skills" of large language models (LLMs). We support everything needed for LLM development, from synthetic data generation, to model training, to evaluation on a wide range of benchmarks. Start developing on a local workstation and move to a large-scale Slurm cluster with just a one-line change.
Here are some of the features we support:
- [Flexible LLM inference](https://nvidia-nemo.github.io/Skills/pipelines/generation/):
- Seamlessly switch between API providers, local server and large-scale slurm jobs for LLM inference.
- Host models (on 1 or many nodes) with [TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM), [vLLM](https://github.com/vllm-project/vllm), [sglang](https://github.com/sgl-project/sglang) or [Megatron](https://github.com/NVIDIA/Megatron-LM).
- Scale SDG jobs from 1 GPU on a local machine all the way to tens of thousands of GPUs on a slurm cluster.
- [Model evaluation](https://nvidia-nemo.github.io/Skills/evaluation):
- Evaluate your models on many popular benchmarks.
- [**Math (natural language**)](https://nvidia-nemo.github.io/Skills/evaluation/natural-math): e.g. [aime24](https://nvidia-nemo.github.io/Skills/evaluation/natural-math/#aime24), [aime25](https://nvidia-nemo.github.io/Skills/evaluation/natural-math/#aime25), [hmmt_feb25](https://nvidia-nemo.github.io/Skills/evaluation/natural-math/#hmmt_feb25)
- [**Math (formal language)**](https://nvidia-nemo.github.io/Skills/evaluation/formal-math): e.g. [minif2f](https://nvidia-nemo.github.io/Skills/evaluation/formal-math/#minif2f), [proofnet](https://nvidia-nemo.github.io/Skills/evaluation/formal-math/#proofnet), [putnam-bench](https://nvidia-nemo.github.io/Skills/evaluation/formal-math/#putnam-bench)
- [**Code**](https://nvidia-nemo.github.io/Skills/evaluation/code): e.g. [swe-bench](https://nvidia-nemo.github.io/Skills/evaluation/code/#swe-bench), [livecodebench](https://nvidia-nemo.github.io/Skills/evaluation/code/#livecodebench), [bird](https://nvidia-nemo.github.io/Skills/evaluation/code/#bird)
- [**Scientific knowledge**](https://nvidia-nemo.github.io/Skills/evaluation/scientific-knowledge): e.g., [hle](https://nvidia-nemo.github.io/Skills/evaluation/scientific-knowledge/#hle), [scicode](https://nvidia-nemo.github.io/Skills/evaluation/scientific-knowledge/#scicode), [gpqa](https://nvidia-nemo.github.io/Skills/evaluation/scientific-knowledge/#gpqa)
- [**Instruction following**](https://nvidia-nemo.github.io/Skills/evaluation/instruction-following): e.g. [ifbench](https://nvidia-nemo.github.io/Skills/evaluation/instruction-following/#ifbench), [ifeval](https://nvidia-nemo.github.io/Skills/evaluation/instruction-following/#ifeval)
- [**Long-context**](https://nvidia-nemo.github.io/Skills/evaluation/long-context): e.g. [ruler](https://nvidia-nemo.github.io/Skills/evaluation/long-context/#ruler), [mrcr](https://nvidia-nemo.github.io/Skills/evaluation/long-context/#mrcr), [aalcr](https://nvidia-nemo.github.io/Skills/evaluation/long-context/#aalcr), [longbench-v2](https://nvidia-nemo.github.io/Skills/evaluation/long-context/#longbench-v2)
- [**Tool-calling**](https://nvidia-nemo.github.io/Skills/evaluation/tool-calling): e.g. [bfcl_v3](https://nvidia-nemo.github.io/Skills/evaluation/tool-calling/#bfcl_v3)
- [**Multilingual**](https://nvidia-nemo.Excerpt of 8,982 characters
Read on GitHub
446
102
101
60
Somshubra Majumdar · NVIDIA · United States
35
34
32
29
27
26
16
14
14
13
13
Sean Naren · @NVIDIA · United Kingdom
11
11
10
10
Hovhannes Tamoyan · @NVIDIA · United States
9
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:a92a96d3eaa9fc2c, llm:Repository description: 'A project to improve skills of large language models' (NVIDIA NeMo project).
matched fp:a92a96d3eaa9fc2c, llm:Repository description: 'A project to improve skills of large language models' (NVIDIA NeMo project).
matched fp:a92a96d3eaa9fc2c, llm:Repository description: 'A project to improve skills of large language models' (NVIDIA NeMo project).