TinyLlama

weights

open(Apache-2.0)

base

TinyLlama-1.1B(open, 3T-token pretrain documented)

post-training-data

open(SFT on UltraChat + DPO on UltraFeedback, both public datasets)

code

open(pretraining project public)

license

Apache-2.0(OSI)

Unusually transparent for a chat model: an open base, and named public alignment datasets (UltraChat for SFT, UltraFeedback for DPO). This category scores what the tuner released - the post-training data - rather than the base pretraining corpus. Zephyr scores fully open on these same two datasets while sitting on Mistral-7B, whose corpus is closed outright, so judging TinyLlama by its base while judging zephyr by its post-training set would penalize the more open of the two bases. One of the most open records in this category.

• https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v1.0 recorded 2026-08-13

  `license:apache-2.0` in the repo tags and `"gated":false`; base TinyLlama-1.1B; SFT on HuggingFaceH4/ultrachat_200k and DPO on HuggingFaceH4/ultrafeedback_binarized, both public; the TinyLlama pretraining project linked from the card