Celso H A Diniz
Celso H A Diniz
I was surprised, it took 45 minutes to complete and the performance was impressive even with this configuration, I only cleared the storage after 20 minutes thinking it was stuck,...
I had the same issue when the dataset was poor in LLaMA models. However, after using a high-quality dataset, I started to notice the same problem across multiple architectures. When...
Here a Notebook exemple before convertion to .GGUF -Base model: --A: https://huggingface.co/meta-llama/Llama-3.2-1B --B: https://huggingface.co/NeuraLakeAi/iSA-02-Nano-1B-Preview/tree/main (finetuned again) -After fine-tuning using high-quality synthetic data, even extending the content to 256K, the model...