Does anyone have an idea of what loss function is used in the fine-tuning process, I have fine-tuned a 4o-mini model with 130 examples and I don’t find an exact documentation that defines the loss function. If I assume its cross-entropy loss, then I can interpret it by a negative exponential function, but how do I confirm this? I don’t want to assume that a full validation loss of 0.5317 is “good”, if I can get the function my interpretation would make sense. Thanks!
Related topics
Topic | Replies | Views | Activity | |
---|---|---|---|---|
Loss Function in Fine Tuning | 3 | 6981 | September 29, 2023 | |
Specific Meanings of Training Loss, Validation Loss, and Full Validation Loss? | 3 | 2515 | August 18, 2024 | |
Finetuning Noob : Guidelines and Best Practices? | 1 | 2507 | September 30, 2023 | |
Loss functions and Optimizer for the funetuning of GPT 3.5 | 0 | 850 | January 11, 2024 | |
Training loss=good, Validation loss=good | 8 | 4078 | April 5, 2024 |