Some concerns about the DeepSeek-Prover-V2 paper. Potentially misformalized examples, and discussion on the Lean zulip suggests the PutnamBench proofs are nonsense and use an implicit sorry (possibly ...