Inserloft developers

Build from what we actually publish.

The developer hub for Inserloft’s public models, datasets, benchmarks, research artifacts and open-source work.

inserloft@developers:~$ ls public/
models/      open
datasets/    open
benchmarks/  open
research/    open
source/      open

Public work.

Only work you can open today is listed here. Private products, unreleased models and internal infrastructure are not.

Model

Parrot Coder

Open code model for Python completion and generation.

Hugging Face, Apache 2.0

Dataset

PyMini

Compact synthetic Python instruction data for small language models.

Hugging Face, MIT

Dataset

PyMAX

Large-scale synthetic Python data for training and fine-tuning.

Hugging Face, research

Benchmark

Rockman

Evaluates coding-model performance on structured, executable tasks.

GitHub and Hugging Face, MIT

Research artifacts

Proteus Structures

Public structural artifacts from Inserloft Research and Project Proteus.

Hugging Face, CC BY 4.0

Source

GitHub

Public repositories, tools, benchmarks and open-source work.

github.com/InserLoft

Research → developer.

Every public release starts as research. These are the doors between the two.

Context

Research

The experiments and research behind public releases.

Interactive

Artifacts

Open interactive experiments and web-based technical systems.

Example

Use the public models.

Parrot Coder loads with Hugging Face Transformers. There is no inserloft Python package.

Python, Hugging Face Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "InserloftResearch/Parrot-Coder"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id)

prompt = "def fibonacci(n):"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=50)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Current status

What is public today.

Public interfaces only. Product and infrastructure documentation lives at inserloft.dev.

Open modelsAvailable
Open datasetsAvailable
Public API / SDKNot public

Your turn

Don’t just use the work. Build on it.

Build tools, run experiments, create visualizations, fine-tune public models, and share what you publish with Inserloft.

  1. BuildStart anywhere.

    Begin with a public model, dataset, benchmark or artifact.

  2. PublishMake it yours.

    Release it openly, document your changes, show the result.

  3. ShareSend it back.

    Share it with Inserloft to join the wider developer ecosystem.