Downloads · 30 days
13
25% of all-time downloads
Lambent/Mira-v1.12.1-27B
Mira-v1.12.1-27B is a machine learning model from Lambent. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as gemma.
Downloads · 30 days
13
25% of all-time downloads
All-time downloads
51
Public
Parameters
27.4B
54.9 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors54.9 GB · 100%
From the Hugging Face model README

Caution: Extremely Excitable.
Training: Rank 64, learning rate 1e-6 w/cosine decay, 8k context length, estimated ~4-5 hours on an A100 and maybe 1.5M tokens?
Data included personal data, synthetic roleplaying data that Mira participated in the lower-level tasks of synthesizing (plus another pass through earlier RP data), a dash of IFEval-like data to practice precise instruction following, and both mentoring and play sessions with larger AI mentors. Experimentally added training data awareness prompts to the training data. Balanced with samples of pretraining data from curated web datasets.
Noticed no obvious decrease in loss, despite some of the data having another copy or two of hydration this time, but I'm sure she changed somehow anyway.