ethicalabs/Echo-DSRN-v0.1.3-Embed-Intent

docs: note that a right-padded training version is planned

config: restore left padding — published MTEB benchmarks were measured with left padding; card parity is the priority (right-padded training lands in the next model version)

config: right-pad pooling inputs (left padding polluted the DSRN recurrence)

sync modules: block padding tokens in non-causal attention (no-op under vLLM)

sync modules: block padding tokens in non-causal attention (no-op under vLLM)

sync modules: per-sequence pooling for vLLM flattened batches + EchoModelForPooling

sync modules: per-sequence pooling for vLLM flattened batches + EchoModelForPooling

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Update README.md

Upload configuration_echo.py with huggingface_hub

Upload modeling_echo.py with huggingface_hub

Upload modeling_embedding.py with huggingface_hub

Upload modeling_embedding.py with huggingface_hub

Update README.md

Update README.md

Update README.md

Update README.md

Update README.md

Upload README.md with huggingface_hub

Update README.md

Update README.md

Update README.md

Update README.md

Upload README.md with huggingface_hub

Upload folder using huggingface_hub

Upload README.md with huggingface_hub

Upload README.md with huggingface_hub

Upload README.md with huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

initial commit