# EPF Cohort 6 Updates – Week 7

My EPF project has taken a shift to align with current developments around Lean Consensus. The project will be pivoting towards [PQdevnet](https://hackmd.io/@reamlabs/S1I-_gHUgg). Following a discussion with Unnawut, it was mutually decided that benchmarking Beacon State's state transition functions would not be as fruitful, and there would be greater value in experimenting with zkVM and [3sf/lean consensus/PQdevnet instead](https://github.com/ReamLabs/ream/pull/672/files).

The goal is that within the next three weeks, I will wrap up my work on [Jolt](https://github.com/x-senpai-x/consenJolt) and [Pico zkvm](https://github.com/x-senpai-x/zkvm-bench/tree/ream-pico/pico) regarding benchmarking the current beacon state's STF. This [repository](https://github.com/x-senpai-x/Ream-ZKVM-Benchmarks) will contain a detailed write-up, benchmarks, learnings, and conclusions for the same, and will compare it with Ream's work on SP1 and RiscZero as well. In parallel, I will start setting up the codebase for benchmarking the new lean state's STFs. Since we currently do not have pyspec tests of blocks on which we could test the transition functions, I will need to mock the beacon state and state transition operations according to the current specifications.

This will have a learning curve; however, I feel that my work and knowledge up until now will be very useful for the upcoming phase of the project.

I have reviewed Vitalik's specifications for 3-slot finality and Ream's current progress with 3sf. I have also studied the [official 3-slot finality post](https://ethresear.ch/t/3-slot-finality-ssf-is-not-about-single-slot/20927) and [the Beam Call for 3sf](https://docs.google.com/presentation/d/1KcvCWEGqjwknS0zgf3yQd8C0Q01CKl84dRKIk4wciEw/edit?slide=id.g335f0c4a985_2_25#slide=id.g335f0c4a985_2_25).

I have attempted to summarize my learnings in this [blog post](https://0xsenpai.substack.com/p/3-slot-finality-the-next-evolution).

Meanwhile, I have also been reading the [complete paper on 3-slot finality](https://arxiv.org/abs/2411.00558), as it genuinely interests me.

Additionally, I received a response from the Pico team regarding the serialization error I was encountering, and it appears I will be able to resolve everything and get the guest running.

## Next Week's Goals

- [x] Read the complete paper on 3sf
- [ ] Develop an initial plan for setting up a zkVM environment for [Lean-consensus](https://github.com/ReamLabs/ream/tree/master/crates/common/consensus/lean)
- [ ] Complete the work on Pico zkvm similar to ConsenJolt