Project Objectives Updated in README

[daveshap]

Outcomes

Here are the primary outcomes we are looking to achieve.

1. Open Source Finetuning Dataset(s): We would like to produce at least one dataset for finetuning models to approximate CoT, MCTS, and Reflection performance.
2. Near-SOTA Benchmarks: We would like to demonstrate near-SOTA performance on reasoning and other benchmarks. We're not trying to achieve SOTA as we are much less funded.
3. Open Source RL-trained Model: Ideally, we'll use reinforcement learning to publish the weights of an open source model.

https://github.com/daveshap/Raspberry/blob/main/README.md