try finetuning encoder-decoder model like t5gemmaHEADmaster

Unfortunately once it finishes the finetuning process, it doesn't
actually output anything. With decoder-only models like small gemma3, it
also doesn't output anything meaningful. I'm tempted to just classify
each token as an ad or not, finetune an encoder like some long-context
BERT, and then a post-processing script to convert back to tokens and
their timestamps. May end up being more performant and efficient too...

1 files changed, 1 insertions, 1 deletions

diff --git a/main.py b/main.py
index a58e02f..6887ac3 100644
--- a/main.py
+++ b/main.py
@@ -401,7 +401,7 @@ def process_transcript_json(input_json_path: str, output_jsonl_path: str):
         
         final_results.append({
             "id": prompt_id,
-            "text": full_prompt_for_finetuning,
+            "input": full_prompt_for_finetuning,
             "target": response_text.strip()
         })