prompt2model
prompt2model copied to clipboard
neulab
prompt2model - Generate Deployable Models from Natural Language Instructions
Some people might be interested in playing around with models that have been trained with prompt2model. Currently, prompt2model supports building gradio demos using the trained models. Here is an example...
Currently prompt2model writes out many files, which can be basically split into two categories: ## Cache These are all of the files that are used by the various models that...
There is another nice library for training models from prompts, [gpt-llm-trainer](https://github.com/mshumer/gpt-llm-trainer) that came out simultaneously with prompt2model. There are some differences in terms of features: | Aspect | prompt2model |...
Currently there are several places where we have ```python try: ... catch Exception: ... ``` blocks. Where possible, it's better to avoid catching something as broad as this. Once we...
Currently, we have benchmarked prompt2model extensively on three tasks (as detailed in our [preprint](https://arxiv.org/abs/2308.12261)). But it would be much cooler if we could benchmark it on a bunch of tasks...
Currently we only use a single retrieved dataset, but we could use multiple retrieved datasets if multiple ones are relevant.
Token indices sequence length is longer than the specified maximum sequence length for this model (973 > 512). Running this sequence through the model will result in indexing errors
@Eren Chenyang Zhao 赵晨阳, of course, you can use `on_epoch_end`, but I had a few concerns at that time. (1) What should I do if I want the `eval_strategy` to...
#34 introduced a prompt parser which uses a long "meta-prompt" to parse a given prompt. Since the "meta-prompt" is long (1.8k tokens), this limits the prompts that can be provided...
In our integration test, we're currently loading a model in the Trainer. this should be mocked to avoid having to download a full transformers model each time this test is...