add vllm and update the world

fixes some cuda lib path problems in vllm
updates mdbook
fixes some clippy lints

later, we should consider splitting the vllm code into a separate set of
deps, so the regular psyche binaries don't get vllm included

Author: arilotter

Cargo.lock

@@ -76,9 +76,9 @@ indicatif = "0.17.5"
 tokenizers = { version = "0.20.0", default-features = false, features = [
   "onig",
 ] }
-tch = { git = "https://github.com/jquesnelle/tch-rs.git", rev = "11d1ca2ef6dbd3f1e5b0986fab0a90fbb6734496" }
-torch-sys = { git = "https://github.com/jquesnelle/tch-rs.git", rev = "11d1ca2ef6dbd3f1e5b0986fab0a90fbb6734496" }
-pyo3-tch = { git = "https://github.com/jquesnelle/tch-rs.git", rev = "11d1ca2ef6dbd3f1e5b0986fab0a90fbb6734496" }
+tch = { git = "https://github.com/jquesnelle/tch-rs.git", rev = "dda507e05a776547a112b6854d1e611684f8c729" }
+torch-sys = { git = "https://github.com/jquesnelle/tch-rs.git", rev = "dda507e05a776547a112b6854d1e611684f8c729" }
+pyo3-tch = { git = "https://github.com/jquesnelle/tch-rs.git", rev = "dda507e05a776547a112b6854d1e611684f8c729" }
 #tch = { path = "../tch-rs" }
 #torch-sys = { path = "../tch-rs/torch-sys" }
 #pyo3-tch = { path = "../tch-rs/pyo3-tch" }

Cargo.toml

@@ -15,9 +15,20 @@ in
     overlays = lib.optionals cudaSupported [ inputs.nix-gl-host.overlays.default ] ++ [
       inputs.rust-overlay.overlays.default
       (final: prev: {
+        # temporary fix until https://github.com/NixOS/nixpkgs/pull/471394 is merged
+        # that lets us use `torch` instead of `torch-bin`
+        cudaPackages = prev.cudaPackages // {
+          cudnn = prev.cudaPackages.cudnn.overrideAttrs (old: {
+            patchelfFlagsArray = (old.patchelfFlagsArray or [ ]) ++ [
+              "--set-rpath"
+              "${prev.lib.getLib prev.cudaPackages.cuda_nvrtc}/lib:\$ORIGIN"
+            ];
+          });
+        };
+
+        # provide packages for uv2pip to include
         python312Packages = prev.python312Packages.override {
           overrides = pyfinal: pyprev: {
-            torch = pyfinal.torch-bin;
             flash-attn = pyfinal.callPackage ../python/flash-attn.nix { };
             liger-kernel = pyfinal.callPackage ../python/liger-kernel.nix { };
             torchtitan = pyfinal.callPackage ../python/torchtitan.nix { };

flake.lock

@@ -1,7 +1,6 @@
 [book]
 authors = ["Nous Research"]
 language = "en"
-multilingual = false
 src = "src"
 title = "Psyche"