Navya-1
Collection
India-first financial LMs from scratch (100M-1.31B), base + SFT chat. navya-1c-sft powers navyam.ai. • 5 items • Updated
navya-1c
Navya is an India-first financial foundation model family, trained from scratch on an India-first corpus (finance + English + Hindi/Hinglish and other Indian languages), custom 64k tokenizer.
| Parameters | 1.31B (dim2048/26L, GQA16/4) |
| Training tokens | ~103B |
| Trained | Aug-Sep 2026 |
| Type | base |
| Author | Navyam AI (Bachatt) |
| License | Apache-2.0 |
| Code | https://github.com/bachatt-app/navyam-gpt |
1.31B from-scratch decoder trained ~103B tokens with a WSD cooldown, on 8xH100. Custom 64k tokenizer.
Intended use: India personal-finance Q&A. Research model — not investment advice.
Usage
from transformers import AutoModelForCausalLM, AutoTokenizer
tok = AutoTokenizer.from_pretrained("navyam-ai/navya-1c")
model = AutoModelForCausalLM.from_pretrained("navyam-ai/navya-1c")
- Downloads last month
- -