| Such a script could look like this (in | |
| pseudocode): | |
| python | |
| model = BrandNewBertModel.load_pretrained_checkpoint("/path/to/checkpoint/") | |
| input_ids = [0, 4, 5, 2, 3, 7, 9] # vector of input ids | |
| original_output = model.predict(input_ids) | |
| Next, regarding the debugging strategy, there are generally a few from which to choose from: | |
| Decompose the original model into many small testable components and run a forward pass on each of those for | |
| verification | |
| Decompose the original model only into the original tokenizer and the original model, run a forward pass on | |
| those, and use intermediate print statements or breakpoints for verification | |
| Again, it is up to you which strategy to choose. |