gemma4a4b: intermediate checkpoints
Intermediate LoRA checkpoints (adapter, optimizer and scheduler state) of the 26B-A4B Ballerina run: gemma4a4b_1-0-0/checkpoint-100 (first 100 steps) and gemma4a4b_1-0-2/checkpoint-500 (the resumed run; final adapter: didula-wso2/gemma4a4b_1-0-2_sft_lora). The expert-layer LoRA uses the training library's (out, experts, rank) column order for lora_B.
Research checkpoint. Licence and usage terms: those of the base model (see the base model card). No warranty; not a product.
- Downloads last month
- -
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for didula-wso2/gemma4a4b_sft_checkpoints
Base model
google/gemma-4-26B-A4B Finetuned
google/gemma-4-26B-A4B-it Finetuned
unsloth/gemma-4-26B-A4B-it