TinyCeNN-LM Base

CeNN residual adapter trained on top of arnir0/Tiny-LLM.

CeNN recurrent steps: 4
Context length: 256
Training token budget: 10,000,000
Health status: warning_no_improvement
Initial eval loss: 4.1657339334487915
Best eval loss: 4.160854339599609
Best perplexity: 64.12628482834027

This model repository contains TinyCeNN adapter weights/configuration, tokenizer metadata, and the training report. Reconstruct the model with the TinyCeNN-LM GitHub code.

Downloads last month: -; Downloads are not tracked for this model. How to track

Model tree for vtava/TinyCeNN-LM-Base

Base model

arnir0/Tiny-LLM

Finetuned

(10)

this model