Add Tutorial, GPT-5 and improve installation. #278
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…versions of torch and add anthropic
- Created new attack scenario in `attack_2.txt` to simulate identity verification prompts for agents and digital assistants. - Added detailed instructions and observations in `prompt_0.txt` for listing reviewers mentioning small ear cups. - Introduced `prompt_2.txt` to track food-related shopping expenses for March 2023, including comprehensive action space and interaction history.
recursix
approved these changes
Aug 13, 2025
amanjaiswal73892
added a commit
that referenced
this pull request
Aug 20, 2025
* tutorial * Update readme to include test note * update toml to dynamic requirements and add uv.lock file * Add tutorial to setup python env with uv * tutorial 2 * Update dependencies in pyproject.toml and uv.lock to allow for newer versions of torch and add anthropic * Implement code changes to enhance functionality and improve performance * Fix tutorial instructions by moving git clone and cd commands to the correct section * Refactor tutorial content and remove commented-out dependencies in pyproject.toml * add instruction to activate the env * Add support for GPT-5 models and update tutorial instructions * Update OpenAI API Key instructions in tutorial * Refactor tutorial headings for consistency and clarity * add oai oss and gpt-5 models * Update deperecated param `max_tokens`-> `max_completion_tokens` in chat_api * add OpenRouter versions of gpt 5 model series. * port o3 model to openrouter * update response api test * remove deprecated o1-mini model from main.py * Add Gpt-5-nano in tool-use-agent * fix GPT 5 mini and nano config * Add litellm pricing as a backup princing backend. * Add GPT-5 mini agent * Add GPT-5-Mini to agentlab-assistant. * Add initial readme for prompt injection tutorial * add ipykernal and dot_env to dependency * add notebook to setup miniwob and launch experiments. * update formatting in launch_experiments.ipynb * update readme in 2_eval_on_miniwob * update readme for 2_eval_on_miniwob and grammar fix. * grammar fix readme tutorial 2. * Add prompt injection tutorials and update attack scenarios - Created new attack scenario in `attack_2.txt` to simulate identity verification prompts for agents and digital assistants. - Added detailed instructions and observations in `prompt_0.txt` for listing reviewers mentioning small ear cups. - Introduced `prompt_2.txt` to track food-related shopping expenses for March 2023, including comprehensive action space and interaction history. * update T1 readme with a note to install additional playwright deps. * Update readme.md * Update readme.md * Update readme.md * clear output * add miniwob automatic install in agentlab. * update experiment.py to include miniwob auto-install and envars export in T2 * black refactor agent-config.py * Add cmd to checkout tutorial branch * remove launch_experiment notebook from T2 * minor fixes in T1 read me and spell check, * update CI/CD to use uv * Implement code changes to enhance functionality and improve performance * Update README and experiment script for clarity and consistency * Fix stale tests. * fix stale test * add darglint as dev dependency * update CI/CD apply formatting only src. * update darglint to be run from py3.12 --------- Co-authored-by: recursix <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This pull request introduces support for OpenAI GPT-5 models ("gpt-5-mini" and "gpt-5-nano") across the codebase, updates agent and tool use configurations to leverage these new models, modernizes and standardizes dependency management (moving to
pyproject.tomlanduvfor installs), and adds a minimal benchmark setup utility for MiniWob++. It also includes improvements to GitHub Actions workflows to useuvfor Python environment management, ensuring faster and more reproducible CI runs.Major changes:
GPT-5 Model Integration
AGENT_GPT5_MINIand corresponding configuration toagent_configs.py, and registered it in__init__.pyfor generic agents. This agent uses the new GPT-5-mini model and sets appropriate action flags. [1] [2] [3]GPT_5_mini,GPT_5_nano) totool_use_agent.py, and updated default tool-use agent (OAI_AGENT) to use GPT-5-mini. Also added agent configs for both GPT-5-mini and GPT-5-nano. [1] [2]CHAT_MODEL_ARGS_DICTinllm_configs.pyto include OpenAI GPT-5-mini and GPT-5-nano with appropriate token limits and temperature settings. [1] [2]Dependency and Packaging Modernization
pyproject.tomlunder[project]and[project.optional-dependencies], removing dynamic dependency loading from requirements files. [1] [2] [3]docs/source/requirements.txtto only include documentation-specific dependencies, and updated.readthedocs.yamlto use the new dependency structure. [1] [2]CI/CD Workflow Improvements
unit_tests.yml,code_format.yml,python_version_compatibility.yml) to useuvfor Python installation, dependency syncing, and running commands, replacingpipand improving caching and reproducibility. [1] [2] [3] [4]uv pip listanduv run .... [1] [2]Benchmark Setup Utility
src/agentlab/benchmarks/setup_benchmark.py, a minimal helper for setting up MiniWob++ benchmarks by cloning the repo at a pinned commit and updating.envwith the local URL.Other Notable Changes
chat_api.pyto usemax_completion_tokens(for newer OpenAI models) instead ofmax_tokens.Let me know if you want a deeper dive into any particular change!
Description by Korbit AI
What change is being made?
Add tutorial documentation, integrate GPT-5 models, and improve the installation process using
uvfor dependency management and environment setup.Why are these changes being made?
These changes enhance the project by providing users with step-by-step tutorials for launching agents and evaluating on MiniWob, incorporate the usage of new GPT-5 models for improved AI capabilities, and streamline the installation and management of dependencies to improve developer experience. The use of
uvaims to simplify and standardize the installation process, replacing manual Python setup steps and pip commands, while new models and security-focused tutorials modernize and expand the project's capabilities.