-
Update test_rag.py
-
Merge branch 'higher_rank' into main
-
finished finetuning with 16 rank lora
-
Fix bug on function evaluation
-
Upload data_augmentation.tsv file (still need to go through queries and check if they are valid)
-
Modify path to relative path
-
Add relative path functionality for colab
-
colab support part two
-
Add support for colab
-
Add a source folder
-
Move paht to a variable
-
Create demo_rag.py
-
Create rag_metadata.py
-
Create test_rag.py
-
Add column evaluator
-
added fine-tune testing on validation set with short queries only
-
Adding validation set testing of newly fine-tuned model
-
switch fine-tuned evaluation script to use 8 rank model
-
trained 8 rank LoRA finetuned model, saving train and validation datasets to disk
-
first cut at fine-tuning with masked training, working but very very slow for inference
-
fixed bug in test_finetuned notebook to read from proper directory
-
Storing 16 and 32 rank LoRA finetuned models
-
Increasing LoRA rank to 32 and updating loss function
-
first cut at adding data augmentation python notebook
-
added python notebook for testing finetuned model
-
First decent attempt, seems to generate valid queries after pre-training. Needs evaluation still
-
First successful fine-tuning run, resulting model is no good but it does train successfully
-
Initial attempt at fine-tuning using LoRA with basic cross-entropy loss
-
ran full testing with updated prompt, improves team and other_stats performance, slight drop on game performance
-
trying to prompt engineer fixes for other_stats table
-
Adding completed pre-training testing runs to python notebook
-
Created evaluation loop for running on full dataframes
-
Merge branch 'main' of https://huggingface.co/USC-Applied-NLP-Group/SQL-Generation into main
-
Updated evaluation function to give tolerance for slight floating point differences
-
train data eda notebook
-
Classify training data
-
Test
-
fixed issue with inconsistent formatting in training data
-
updating prompt to include team name and abbreviations
-
Small tweaks to evaluation function
-
updating test pretrained script to evaluate model outputs
-
Renamed demo script and added initial pre-trained test python notebook
-
Adding SQL-NLP training data
-
added better prompt engineering
-
tweaking prompt a little further so it only outputs SQL more consistently
-
Updated to use real nba database for test query
-
added nba kaggle sqlite database and testing python notebook
-
Update README.md
-
Create README.md
-
Delete README.md
