aleada/Gemma-3-12B-it-W4A16

model card: ASSERT header, and the two companion tools the card did not mention

model card: the figure the recovery section references

model card: measured recovery against the source model

model card: assert-logo-dark.png, referenced by the card

model card: assert-logo-light.png, referenced by the card

model card: ASSERT header, and the two companion tools the card did not mention

add a link to the pack integrity checker

model card: record the dampening the run actually used

model card: drop a rationale copied from a different model

docs(card): vLLM-loading-fixed banner + honest changelog (transformers>4.52.2 naming; byte-identical weights)

vLLM-naming fix (model.safetensors.index.json): remap multimodal keys to vLLM convention; weights byte-identical, no re-quant

vLLM-naming fix (model-00002-of-00002.safetensors): remap multimodal keys to vLLM convention; weights byte-identical, no re-quant

vLLM-naming fix (model-00001-of-00002.safetensors): remap multimodal keys to vLLM convention; weights byte-identical, no re-quant

fix(card): substitute placeholders + add vLLM example

fix(card): substitute placeholders + add vLLM example

Add tokenizer_config.json

Add tokenizer.json

Add quantize_config.json

Add quant_log.csv

Add sharded W4A16 safetensors via git-lfs

Add generation_config.json

Add config.json

Add chat_template.jinja

Add README.md

initial commit