it-just-works/stella_en_1.5B_v5_bf16

docs: document fork provenance and transformers 5.x / 4.45+ compatibility fixes

fix: rebuild rotary caches lazily for transformers >= 5 meta-device init

Add: convert_to_bf16.py script

fix: issues with latest transformers, add previously removed function to compute usable past KV length for cache compatibility

Update: del onnx variants

Update README.md

Update README.md

Upload ONNX weights (#3)

Add infinity to readme (#39)

add infinity example in the readme (#32)

Update README.md

Making File Compatible With Environments That Do Not Have Flash Attention (#26)

Adding `safetensors` variant of this model (#2)

Fix query_prompt_name variable name (#15)

Set 1024 as default dim, update usage snippets, store prompts in config (#1)

Upload README.md

Upload 17 files

Upload README.md

Upload README.md

Upload 12 files

initial commit