QagentS commited on
Commit
de9b3e2
1 Parent(s): 2c6c4a6

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +5 -0
README.md CHANGED
@@ -33,9 +33,14 @@ widget:
33
  A 1.3 bn SQL model that outperforms most SQL expert models and chatgpt on popular benchmarks.
34
  This is a distilled model built on the deepseek base model.
35
 
 
 
36
  ## How we built it?
37
 
38
  We used softmax cross entropy and a modified form of policy grad along with Q loss, optimized in an EM set up.
 
 
 
39
 
40
  ## Benchmarking :
41
  For benchmarking purposes we are using Semantic Evaluation for Text-to-SQL with
 
33
  A 1.3 bn SQL model that outperforms most SQL expert models and chatgpt on popular benchmarks.
34
  This is a distilled model built on the deepseek base model.
35
 
36
+
37
+
38
  ## How we built it?
39
 
40
  We used softmax cross entropy and a modified form of policy grad along with Q loss, optimized in an EM set up.
41
+ Loss behaviour in the set up mentioned above -
42
+
43
+ ![image/png](https://cdn-uploads.huggingface.co/production/uploads/658d8095a2a6a6e0da8bb8a6/I80Ru1r4thoYrLagIWALa.png)
44
 
45
  ## Benchmarking :
46
  For benchmarking purposes we are using Semantic Evaluation for Text-to-SQL with