Logistic Regression MVP #2

arihantbansal · 2024-12-21T06:08:06Z

Possible improvements

Array inputs for coeff, and input
Import existing model weights using some lib and showcase in test (?)

cryptopapi997 · 2024-12-21T14:46:03Z

predictor/README.md

+This project is structured pretty similarly to how a regular Solana Anchor project is structured. The main difference lies in there being two places to write code here:
+
+- The `programs` dir like normal
+- The `confidential-ixs` dir for confidential computing instructions
+
+When working with plaintext data, we can edit it inside our program as normal. When working with confidential data though, state transitions take place off-chain using the Arcium network as a co-processor. For this, we then always need two instructions in our program: one that gets called to initialize a confidential computation, and one that gets called when the computation is done and supplies the resulting data. Additionally, since the types and operations in a Solana program and in a confidential computing environment are a bit different, we define the operations themselves in the `confidential-ixs` dir using our Rust-based framework called Arcis. To link all of this together, we provide a few macros that take care of ensuring the correct accounts and data are passed for the specific initialization and callback functions:
+
+```
+// confidential-ixs/add_together.rs
+
+use arcis::prelude::*;
+
+arcis_main!();
+
+// mu8 is a masked u8, i.e. an encrypted u8.
+#[circuit]
+fn add_together(x: mu8, y: mu8) -> mu8 {
+    x + y
+}
+
+// programs/my_program/src/lib.rs
+
+use anchor_lang::prelude::*;
+use arcium_anchor::queue_computation;
+use arcium_macros::{arcium_callback, callback_accounts, queue_computation_accounts};
+
+declare_id!("<some ID>");
+
+#[program]
+pub mod my_program {
+    use super::*;
+
+    pub fn init_computation(_ctx: Context<InitComputation>, input: Vec<u8>) -> Result<()> {
+        queue_computation(_ctx.accounts, input, vec![])?;
+        Ok(())
+    }
+
+    #[arcium_callback(circuit = "add_together")]
+    pub fn add_together_callback(ctx: Context<Callback>, output: Vec<u8>) -> Result<()> {
+        msg!("Arcium callback invoked with output {:?}", output);
+        Ok(())
+    }
+}
+
+#[callback_accounts(circuit = "add_together")]
+pub struct Callback<'info> {
+    pub some_extra_acc: AccountInfo<'info>,
+}
+
+#[queue_computation_accounts(circuit = "add_together")]
+pub struct InitComputation<'info> {
+    pub some_extra_acc: AccountInfo<'info>,
+}
+```


This is quite general - a few words about what this sample is about would be better here

predictor/programs/predictor/src/lib.rs

predictor/tests/predictor.ts

arihantbansal added 2 commits December 21, 2024 11:34

mvp

9e34093

rename

5fd9bed

cryptopapi997 requested changes Dec 21, 2024

View reviewed changes

arihantbansal added 2 commits December 21, 2024 21:38

wip

e62344e

wip

1c5c258

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Logistic Regression MVP #2

Logistic Regression MVP #2

arihantbansal commented Dec 21, 2024

cryptopapi997 Dec 21, 2024

Logistic Regression MVP #2

Are you sure you want to change the base?

Logistic Regression MVP #2

Conversation

arihantbansal commented Dec 21, 2024

cryptopapi997 Dec 21, 2024

Choose a reason for hiding this comment