Dusting the project off

.typos.toml

@@ -1,4 +1,5 @@
 [default]
 extend-ignore-identifiers-re = [
     "mmaped",
+    "arange",
 ]

README.md

@@ -3,14 +3,9 @@
 </p>
 
 [![Continuous integration](https://github.com/EricLBuehler/candle-vllm/actions/workflows/ci.yml/badge.svg)](https://github.com/EricLBuehler/candle-vllm/actions/workflows/ci.yml)
-[![Discord server](https://dcbadge.vercel.app/api/server/FAeJRRJ8)](https://discord.gg/FAeJRRJ8)
 
 Efficient, easy-to-use platform for inference and serving local LLMs including an OpenAI compatible API server.
 
-PPlease see [mistral.rs](https://github.com/EricLBuehler/mistral.rs), efficient inference platform for many models, including quantized support. Additionally, it implements X-LoRA, recently released method [here](https://github.com/EricLBuehler/xlora). X-LoRA introduces a MoE inspired method to densely gate LoRA adapters powered by a model self-reflection forward pass.
-
-**candle-vllm is flux, in breaking development and as such is currently unstable.**
-
 ## Features
 - OpenAI compatible API server provided for serving LLMs.
 - Highly extensible trait-based system to allow rapid implementation of new module pipelines,
@@ -23,8 +18,6 @@ PPlease see [mistral.rs](https://github.com/EricLBuehler/mistral.rs), efficient
     - 7b
     - 13b
     - 70b
-- Mistral
-    - 7b
 
 ## Examples
 See [this folder](examples/) for some examples.

kernels/build.rs

@@ -1,8 +1,8 @@
-use std::path::PathBuf;
+use anyhow::Result;
+use std::fs::read_to_string;
 use std::fs::OpenOptions;
 use std::io::prelude::*;
-use anyhow::{Result};
-use std::fs::read_to_string;
+use std::path::PathBuf;
 
 fn read_lines(filename: &str) -> Vec<String> {
     let mut result = Vec::new();
@@ -29,27 +29,27 @@ fn main() -> Result<()> {
     let kernel_dir = PathBuf::from("../kernels/");
     let absolute_kernel_dir = std::fs::canonicalize(&kernel_dir).unwrap();
 
-    println!("cargo:rustc-link-search=native={}", absolute_kernel_dir.display());
-    println!("cargo:rustc-link-lib=pagedattention"); 
+    println!(
+        "cargo:rustc-link-search=native={}",
+        absolute_kernel_dir.display()
+    );
+    println!("cargo:rustc-link-lib=pagedattention");
     println!("cargo:rustc-link-lib=dylib=cudart");
-    
+
     let contents = read_lines("src/lib.rs");
     for line in contents {
         if line == "pub mod ffi;" {
-            return Ok(())
+            return Ok(());
         }
     }
     let mut file = OpenOptions::new()
-                    .write(true)
-                    .append(true)
-                    .open("src/lib.rs")
-                    .unwrap();
+        .write(true)
+        .append(true)
+        .open("src/lib.rs")
+        .unwrap();
     //Expose paged attention interface to Rust
     if let Err(e) = writeln!(file, "pub mod ffi;") {
-        anyhow::bail!(
-            "error while building dependencies: {:?}\n",
-            e,
-        )
+        anyhow::bail!("error while building dependencies: {:?}\n", e,)
     } else {
         Ok(())
     }

kernels/src/ffi.rs

@@ -67,4 +67,4 @@ extern "C" {
 
         dtype: u32,
     );
-}
+}

kernels/src/lib.rs

@@ -1,4 +1,6 @@
-pub const COPY_BLOCKS_KERNEL: &str = include_str!(concat!(env!("OUT_DIR"), "/copy_blocks_kernel.ptx"));
+pub const COPY_BLOCKS_KERNEL: &str =
+    include_str!(concat!(env!("OUT_DIR"), "/copy_blocks_kernel.ptx"));
 pub const PAGEDATTENTION: &str = include_str!(concat!(env!("OUT_DIR"), "/pagedattention.ptx"));
-pub const RESHAPE_AND_CACHE_KERNEL: &str = include_str!(concat!(env!("OUT_DIR"), "/reshape_and_cache_kernel.ptx"));
+pub const RESHAPE_AND_CACHE_KERNEL: &str =
+    include_str!(concat!(env!("OUT_DIR"), "/reshape_and_cache_kernel.ptx"));
 pub mod ffi;