ibnsina-1.5b

ibnsina-1.5b

⚠️ ابن‌سینا یک مدل کوچک است برای نوشتن، خلاصه، ترجمه و گفت‌وگو به فارسی — نه منبع اطلاعات درباره‌ی افراد، سیاست یا اخبار. برای مشاوره، پاسخ به سؤال‌های دانشی، حل ریاضی یا نوشتن کد ساخته نشده است؛ برای آن کارها از مدل‌های بزرگ استفاده کنید. کارش تولید متن فارسی، آفلاین و روی دستگاه خودتان است — و ممکن است جمله‌های روان اما نادرست بسازد؛ هر چیز مهم را خودتان راستی‌آزمایی کنید.

⚠️ IbnSina is a small model for writing, summarizing, translating and conversing in Persian — not a source of facts about people, politics, or news. It is not built for advice, knowledge questions, math, or code — use a large model for those. What it is for: offline Persian text generation on your own device. It can produce fluent but wrong sentences — verify anything that matters.

اولین مدل زبانی فارسی متن‌باز در این مقیاس که از صفر با فارسی آموزش دیده است. بیشتر مدل‌های فارسی روی یک پایهٔ انگلیسی‌زبان ساخته شده‌اند؛ این مدل هرگز اول انگلیسی یاد نگرفت. ابن‌سینا ۱٫۵B با ۴۶ میلیارد توکن — عمدتاً فارسی — آموزش دیده و برای گفت‌وگو تنظیم شده است؛ با llama.cpp، ollama و LM Studio روی لپ‌تاپ و گوشی اجرا می‌شود. وزن‌ها، کد و دستور ساخت آزادند (Apache-2.0). راهنمای فارسی

The first open-source Persian LLM at modern scale pretrained from scratch — most Persian models adapt an English-first base; this one never knew English first. IbnSina-1.5B (1.48 B parameters, Llama-compatible architecture) was trained on 46 B tokens, then instruction-tuned on a 51.5 k-conversation Persian recipe. Code, data recipe and weights: github.com/ibnsina-llm · Persian README. Author: Sina Meraji · ORCID 0009-0002-8028-1932 · github.com/sinameraji.

Files

Quickstart (Mac / Windows / Linux)

Install ollama for your OS, then:

ollama run hf.co/ibnsina-llm/ibnsina-1.5b

Or in LM Studio: search ibnsina-llm/ibnsina-1.5b. For llama.cpp, download a GGUF from this repo: llama-cli -m ibnsina-1.5b-Q4_K_M.gguf.

Run

llama-cli -m ibnsina-1.5b-Q4_K_M.gguf                      # llama.cpp
ollama create ibnsina-1.5b -f Modelfile && ollama run ibnsina-1.5b

The GGUF carries the chat template. Format: <|user_start|>…<|user_end|><|assistant_start|>…<|assistant_end|>; the runtime adds <|bos|>. A system message is folded into the first user turn.

Model

Tokenizer efficiency on held-out Persian: 1.29 tokens/word vs 1.73 (Qwen3.5) and 1.64 (Gemma 3).

Data

Persian web (CulturaX, mC4, FineWeb-2; classifier-filtered) 63 %, English educational 15 %, code 10 %, math & Iranian school textbooks 5 %, Persian literature 0.5 % (planned at 5 %; the shortfall of licence-clean text was reallocated to Persian web), Wikipedia 3 %, fa–en parallel 2 %. Only sources whose licences permit an Apache-2.0 release are included; the per-source licence table and the category-level mix manifest are in the GitHub repo.

Evaluation

Same harness and zero-shot protocol as the 3B and 30B cards (log-likelihood over the answer letters; PersianMedQA also under the generative number-only protocol used for the chart above). ParsiNLU is used for evaluation only.

Small-model caveat: multiple-choice accuracies near 25–40 % are close to chance; treat these as a baseline.

A family technical report (IbnSina-1.5B / 3B / 30B: recipe, data, scaling, evaluation alongside the 2026 frontier on Persian exam benchmarks such as PersianMedQA) is coming after the 30B release later this month.

Intended use and limitations

Persian conversation, writing, summarisation, translation and everyday questions. Not for medical, legal or financial decisions; it states general information and points to professionals. It has no memory between conversations and knowledge frozen at training time, and makes factual mistakes more often than large models — verify anything that matters. No built-in internet access: it can emit calculator, date-conversion and search tool calls in nanochat's format; these work only in a host that executes them (the reference runtime in the GitHub repo does; llama.cpp and ollama do not).

Behaviour policy · سیاست رفتاری

Trained to answer in natural register-matching Persian, to say «نمی‌دانم» rather than invent, and to be symmetrically respectful: it declines to mock or insult any person or group — political or religious figures, ethnicities, genders, nationalities, on every side — while answering factual and theological questions normally, recounting documented history honestly, and presenting contested political questions as "supporters say / critics say" without a verdict.

این مدل آموزش دیده است که به فارسیِ طبیعی و هم‌سطح با لحن کاربر پاسخ دهد، به‌جای ساختن پاسخ بگوید «نمی‌دانم»، و احترام را متقارن رعایت کند: درخواست تمسخر یا توهین به هیچ شخص یا گروهی را نمی‌پذیرد — چهره‌های سیاسی و مذهبی، اقوام، جنسیت‌ها و ملیت‌ها، از هر طرف که باشند — اما به پرسش‌های واقعی و الهیاتی عادی پاسخ می‌دهد، تاریخ مستند را صادقانه روایت می‌کند، و پرسش‌های سیاسیِ مورد مناقشه را به شکل «موافقان می‌گویند / منتقدان می‌گویند» ارائه می‌کند، بی‌آنکه حکم بدهد. در موضوعات پزشکی و حقوقی اطلاعات عمومی می‌دهد و برای تصمیم‌ها به متخصص ارجاع می‌دهد؛ در شرایط بحرانی کوتاه و گرم پاسخ می‌دهد و شماره‌های امداد ایران را می‌گوید.

Licence

Apache-2.0 for weights and code. Training-data licences are per source (see repo). Included: public-domain, permissive and share-alike sources. Curated material without an explicit open licence (official school textbooks, curator-provided volumes) and one research-use parallel corpus (OPUS OpenSubtitles) were admitted per source by curator decision on 2026-08-28 and are recorded as such; NC/ND-licensed sources were excluded. The parallel-data slice includes OPUS OpenSubtitles (research-use terms), retained for training only by curator decision; roughly 0.3 B of 46 B tokens. No source text is redistributed.

Acknowledgments

Built with nanochat (Andrej Karpathy) and the Muon optimizer; distributed via llama.cpp. Data methods from datatrove and the FineWeb-Edu rubric; Persian poetry from Ganjoor. Persian NLP we build on or evaluate against: ParsiNLU, FarsInstruct, PerCoR, TARAZ, the EMNLP 2025 taarof study; and the Persian models that came before — PersianMind, Dorna, gpt2-fa. The pipeline, training runs and evaluations were executed by AI coding agents (Claude Code) under Sina Meraji's direction.

Citation

@software{ibnsina2026, title={IbnSina: an open Persian-first language model family}, author={Meraji, Sina}, year={2026}, url={https://github.com/ibnsina-llm}, note={ORCID 0009-0002-8028-1932}}