Add support for saving intermediate results during vf-eval

[bsagevedant]

Save Intermediate Results During vf-eval

This PR addresses issue #251 by adding support for saving intermediate results during evaluation and enabling interleaved reward computation.

Changes

• Added configuration options to Environment class:

  • save_intermediate: Enable saving intermediate results during rollout
  • interleave_rewards: Enable computing rewards after each rollout instead of batching

• Modified run_rollouts method to:

  • Support saving intermediate results after each rollout
  • Support interleaving reward computation
  • Make both features optional and configurable

• Added comprehensive tests in test_intermediate_results.py

Testing

Added new test cases that verify:

• Intermediate results saving functionality
• Interleaved reward computation
• Configuration options
• Integration with existing evaluation methods

Notes

• The interleaved reward computation is optional as it's not fully compatible with some pairwise reward strategies
• Intermediate results are logged using the environment's logger, which can be customized by the user

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
1 out of 2 committers have signed the CLA.

✅ willccbb
❌ Your GitHub Username

---

Your GitHub Username seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[willccbb]

nice! looks pretty good, updated to merge with latest main -- probably will make some other edits before merging, our logic for vf-eval outputs json saving has drifted a bit from make_dataset + ideally we bring these back in sync so that intermediate saving would handle vf-eval -s directly.