update generation examples and params

README.md

@@ -1,9 +1,10 @@
 ---
 license: apache-2.0
 tags:
-- generated_from_trainer
-model-index:
-- name: t5-v1_1-base-ft-jflAUG
+- grammar
+- spelling
+- punctuation
+- error-correction
 
 widget:
 - text: "Anna and Mike is going skiing"
@@ -26,24 +27,27 @@ ta ohow to remove trents in these nalitives from time series"
   example_title: "dangling modifier"
 - text: "There car broke down so their hitching a ride to they're class."
   example_title: "compound-1"
+- text: "Which part of Zurich was you going to go hiking in when we were there for the first time together? ! ?"
+  example_title: "chatbot on Zurich"
 
 inference:
   parameters:
-    no_repeat_ngram_size: 2
+    no_repeat_ngram_size: 4
     max_length: 64
     min_length: 4
     num_beams: 4
-    repetition_penalty: 3.51
-    length_penalty: 0.8
+    repetition_penalty: 1.51
+    length_penalty: 1
     early_stopping: True
 ---
 
 
 # t5-v1_1-base-ft-jflAUG
 
-- **GOAL:** build a more robust and generalized grammar and spelling correction model with minimal impact on the semantics of correct sentences (I.e. it does not change things that do not need to be changed.
-- this grammar correction model (at least from preliminary testing) can handle large amounts of errors in the source text (i.e. from audio transcription) and still produce cohesive results. 
-- This model is a fine-tuned version of [google/t5-v1_1-base](https://huggingface.co/google/t5-v1_1-base) on an expanded version of the [JFLEG dataset](https://aclanthology.org/E17-2037/).
+> **GOAL:** a more robust and generalized grammar and spelling correction model with minimal impact on the semantics of correct sentences (i.e. it does not change things that do not need to be changed).
+
+- this model _(at least from preliminary testing)_ can handle large amounts of errors in the source text (i.e. from audio transcription) and still produce cohesive results. 
+- a fine-tuned version of [google/t5-v1_1-base](https://huggingface.co/google/t5-v1_1-base) on an expanded version of the [JFLEG dataset](https://aclanthology.org/E17-2037/).
 
 ## Model description
 
@@ -58,9 +62,8 @@ inference:
 
 ## Training and evaluation data
 
-More information needed
-
-## Training procedure
+- trained as text-to-text
+- JFLEG dataset + additional selected and/or generated grammar corrections
 
 ### Training hyperparameters
 
@@ -77,9 +80,6 @@ The following hyperparameters were used during training:
 - lr_scheduler_warmup_ratio: 0.05
 - num_epochs: 5
 
-### Training results
-
-
 
 ### Framework versions