Upload README.md

README.md

@@ -1,12 +1,9 @@
----
-license: cc-by-4.0
----
 # tokenspace directory
 
 This directory contains utilities for the purpose of browsing the
 "token space" of CLIP ViT-L/14
 
-Long term goal is to be able to literally browse from word to
+Long term goal is to be able to literally browse from word to 
 "nearby" word
 
 
@@ -15,15 +12,29 @@ Long term goal is to be able to literally browse from word to
 
 Generates the "embeddings.safetensor" file!
 
-Basically goes through the fullword.json file, and
+Basically goes through the fullword.json file, and 
 generates a standalone embedding object for each word.
 Shape of the embeddings tensor, is
  [number-of-words][768]
 
+Note that it is possible to directly pull a tensor from the CLIP model,
+key of  text_model.embeddings.token_embedding.weight
+
+This will NOT GIVE YOU THE RIGHT DISTANCES!
+Hence why we are calculating and then storing the embedding weights actually
+used by the CLIP process
+
+
+## generate-distances.py
+
+Loads the prior generated embeddings, and then tries to calculate a full matrix
+of distances between all tokens
+
+
 ## fullword.json
 
 This file contains a collection of "one word, one CLIP token id" pairings.
 The file was taken from vocab.sdturbo.json
-First all the non-</w> entries were stripped out.
+First all the non-(/w) entries were stripped out.
 Then all the garbage punctuation and foreign characters were stripped out.
-Finally, the actual </w> was stripped out, for ease of use.
+Finally, the actual (/w) was stripped out, for ease of use.