Edsl

Added
- `Results` method `spot_issues()` runs a survey to spot issues and suggest revised versions of any prompts that did not generate responses in your original survey (i.e., any user/system prompts where your results show a null answer and raw model response). You can optionally pass a list of models to use to run the meta-survey instead of the default model. See details on the meta-questions that are used and how it works: https://www.expectedparrot.com/content/RobinHorton/spot-issues-notebook.

- When you post an object to Coop with the `push()` method you can optionally pass a `description`, a convenient `alias` for the Coop URL that is created and a `visiblity` setting (*public*, *private* or *unlisted* by default). An alias Coop URL is now displayed in the object details that are returned when the object is created. You can then use the `alias_url` to retrieve or modify the object in lieu of the `uuid`. See examples in the [Coop section](https://docs.expectedparrot.com/en/latest/coop.html).

- `Scenario` objects can be reference with the `scenario.` prefix, e.g., "Do you enjoy {{ scenario.activity }}?" (previously "Do you enjoy {{ activity }}?") to standardize syntax with other objects, e.g., when referencing `agent.` fields in the same way, or when piping `answer.` and `question.` fields.

Added
- A universal remote cache (URC) is available for retrieving responses to any questions that have been run at the Expected Parrot server. If you re-run a question that anyone has run before, you can retrieve that response at no cost to you. This cache is available for all jobs run remotely by default, and new responses are automatically added to it. If you want to draw fresh responses you can use `run(fresh=True)`. If you draw a fresh response for a question that has already been run, the new response is also added to the URC with an iteration index. The URC is not available for jobs run locally. See the [remote cache](https://docs.expectedparrot.com/en/latest/remote_caching.html) section for details and FAQ.

- `ScenarioList` methods for concatenating and collapsing scenarios in a scenario list:

* `concatenate()` can be used to concatenate specified fields into a single string field

* `concatenate_to_list()` can be used to concatenate specified fields into a single list field

* `concatenate_to_set()` can be used to concatenate specified fields into a single set field

* `collapse()` can be used to collapse a scenario list by grouping on all fields except a specified field

See [examples](https://docs.expectedparrot.com/en/latest/scenarios.html#combining-scenarios).

- `ScenarioList` method `from_sqlite()` can be used to create a scenario list from a SQLite database.

Fixed
- Bug causing some tokens generated to be omitted from results when skip logic was applied.

Added
- `ScenarioList` method `from_dta()` creates a scenario list from a Stata file.

- `Results` method `flatten()` will flatten a field of dictionaries into separate fields. It takes a list of the fields to flatten and a boolean indicator whether to preserve the original fields in the new `Results` object that is returned. [See example](https://docs.expectedparrot.com/en/latest/results.html#flattening-resuls).

- `Results` method `report()` generates a report of selected columns in markdown by iterating through the rows, presented as observations. You can optionally pass headers, a divider and a limit on the number of observations to include. It can be useful if you want to display some sample part of larger results in a working notebook you are sharing. [See example](https://docs.expectedparrot.com/en/latest/results.html#generating-a-report).

- `Survey` method `show_flow()` can now also be called on a `Jobs` object, and will show any scenarios and/or agent traits that that you have added to questions. [See examples](https://docs.expectedparrot.com/en/latest/docs/surveys.html#show-flow).

Added
- Xai models are now available. If you have your own key, you can add it to your Keys page at your Coop account or add `XAI_API_KEY=<your_key_here>` to your `.env` file.

- `Survey` method `humanize()` will create a web-based version of your survey to share with humans. Responses are automatically added to a `Results` object that you can access at your account. *This feature is live but in development.*

Added
- You can now use your own keys from service providers to run jobs remotely at the Expected Parrot server, and store them at the [Keys](https://www.expectedparrot.com/home/keys) page of your [Coop account](https://www.expectedparrot.com/login) (in lieu of your `.env` file). You can also grant access to other users (without sharing the keys directly), set limits on their usage and set RPM/TPM limits.

- You can run a remote survey in the background (and then continue working or not) by calling `run(background=True)`. You can check the status of the job at any time by (1) viewing the progress bar page (the link is returned while your job is running), (2) calling `results.fetch()` (which will return a status update every 1.0 seconds or the `polling_interval` that you specify, or the completed results) or (3) calling the results as usual, e.g., `results.columns`. Additional planned features: request email notification when your job is completed. See an [example](https://docs.expectedparrot.com/en/latest/run_background.html).

- Method `ScenarioList.from_pdf_to_image(<filename>)` generates a scenario for each page of a pdf converted into a jpeg (to use as an image instead of converting to text). Companion method `Scenario.from_pdf_to_image(<filename>)` generates a key/value for each page within the same scenario object to allow you to use multiple images at the same time. See a [notebook of examples](https://www.expectedparrot.com/content/ea777fab-9cb1-4738-8fa3-bbdef20ed60d).

- You can now pull an object from Coop using its alias. Alias routes were previously of the form expectedparrot.com/<owner_username>/<alias>, They are now of the form expectedparrot.com/content/<owner_username>/<alias>.

- You can now see Mermaid diagrams and inline math in Coop notebooks.

Fixed
- Improved methods and moved tasks to background to prevent some timeout errors.

- Upgrated connection for Anthropic models.

- Fixed a bug preventing iterations on remote inference.

Added
- DeepSeek models are now available (e.g., try `Model("deepseek-reasoner")`). If you have your own key, you can add it to your Keys page at your Coop account or add `DEEPSEEK_API_KEY=<your_key_here>` to your `.env` file.

- The name of the inference service is now included in the `Model` parameters and `Results` objects. This can be useful when the same model is provided by multiple services.

- The model pricing page at Coop shows daily test results for available models: https://www.expectedparrot.com/home/pricing. The same information can also be returned by calling the method `Model.check_working_models()`. Check the models for a particular service provider by passing the name of the service: `Model.check_working_models(service="google")`.

Changed
- Default size limits on question texts have been removed.