Upload folder using huggingface_hub

README.md

@@ -87,6 +87,10 @@ language:
 - yo
 - zh
 ---
+# Accuracy
+
+Not sure on accuracy quite yet, will update soon. After I confirm this is working well (preliminary results suggest it's good), I can try a version which combines normalization + quantization for the `token_embeddings` output.
+
 # snowflake2_m_uint8
 
 This is a slightly modified version of the uint8 quantized ONNX model from https://huggingface.co/Snowflake/snowflake-arctic-embed-m-v2.0
@@ -99,7 +103,7 @@ No benchmarks, but it in my limited testing it's exactly equivalent to the FP32
 
 # Quantization method
 
-Determined value range for FP32 tensor is -.31 to 0.31.  I quantized according to that scale directly in this ONNX model.
+Linear quantization for the scale -.3 to 0.3, which is what sentence_embedding is normalized to.
 
 Here's what the graph of the original output looks like:
 

snowflake2_m_uint8.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0ef454690f03d5b63dc333b2cb41611fa26d64a3efe7e0b21947cb953cce2a43
-size 310916367
+oid sha256:97de99227c030abf5207feb4e8cb75a65caa43dae4df3a1defb8ec8743c70b8b
+size 310916368