SQL-Generation / finetune_model.ipynb

Commit History

fix cuda

ef32c53

licesma commited on 11 days ago

Fix colab bugs

9687f25

licesma commited on 11 days ago

Prepare fine-tune for colab

569052e

licesma commited on 11 days ago

finished finetuning with 16 rank lora

3ad5864

DeanGumas commited on 11 days ago

added fine-tune testing on validation set with short queries only

f94c0b8

DeanGumas commited on 13 days ago

Adding validation set testing of newly fine-tuned model

98e9ca0

DeanGumas commited on 13 days ago

trained 8 rank LoRA finetuned model, saving train and validation datasets to disk

c72582f

DeanGumas commited on 14 days ago

first cut at fine-tuning with masked training, working but very very slow for inference

a30a6cf

DeanGumas commited on 15 days ago

Increasing LoRA rank to 32 and updating loss function

c78bae0

DeanGumas commited on 15 days ago

First decent attempt, seems to generate valid queries after pre-training. Needs evaluation still

fdaf162

DeanGumas commited on 18 days ago

First successful fine-tuning run, resulting model is no good but it does train successfully

1f9c86c

DeanGumas commited on 19 days ago

Initial attempt at fine-tuning using LoRA with basic cross-entropy loss

8c47142

DeanGumas commited on 19 days ago

Commit History

fix cuda ef32c53

Fix colab bugs 9687f25

Prepare fine-tune for colab 569052e

finished finetuning with 16 rank lora 3ad5864

added fine-tune testing on validation set with short queries only f94c0b8

Adding validation set testing of newly fine-tuned model 98e9ca0

trained 8 rank LoRA finetuned model, saving train and validation datasets to disk c72582f

first cut at fine-tuning with masked training, working but very very slow for inference a30a6cf

Increasing LoRA rank to 32 and updating loss function c78bae0

First decent attempt, seems to generate valid queries after pre-training. Needs evaluation still fdaf162

First successful fine-tuning run, resulting model is no good but it does train successfully 1f9c86c

Initial attempt at fine-tuning using LoRA with basic cross-entropy loss 8c47142

fix cuda

ef32c53

Fix colab bugs

9687f25

Prepare fine-tune for colab

569052e

finished finetuning with 16 rank lora

3ad5864

added fine-tune testing on validation set with short queries only

f94c0b8

Adding validation set testing of newly fine-tuned model

98e9ca0

trained 8 rank LoRA finetuned model, saving train and validation datasets to disk

c72582f

first cut at fine-tuning with masked training, working but very very slow for inference

a30a6cf

Increasing LoRA rank to 32 and updating loss function

c78bae0

First decent attempt, seems to generate valid queries after pre-training. Needs evaluation still

fdaf162

First successful fine-tuning run, resulting model is no good but it does train successfully

1f9c86c

Initial attempt at fine-tuning using LoRA with basic cross-entropy loss

8c47142