Llama more training (#297)

* Rework the var-builder to handle initializations.

* Add some helper functions for layer creation.

* Improve the layer initializations.

* Get initialized variables.

* Precompute the rot embeddings when training lamas.

candle-nn/src/embedding.rs

@@ -28,3 +28,15 @@ impl Embedding {
         Ok(values)
     }
 }
+
+pub fn embedding(in_size: usize, out_size: usize, vb: crate::VarBuilder) -> Result<Embedding> {
+    let embeddings = vb.get_or_init(
+        (in_size, out_size),
+        "weight",
+        crate::Init::Randn {
+            mean: 0.,
+            stdev: 1.,
+        },
+    )?;
+    Ok(Embedding::new(embeddings, out_size))
+}

candle-nn/src/layer_norm.rs

@@ -62,3 +62,9 @@ impl LayerNorm {
         Ok(x)
     }
 }
+
+pub fn layer_norm(size: usize, eps: f64, vb: crate::VarBuilder) -> Result<LayerNorm> {
+    let weight = vb.get_or_init(size, "weight", crate::Init::Const(1.))?;
+    let bias = vb.get_or_init(size, "bias", crate::Init::Const(0.))?;
+    Ok(LayerNorm::new(weight, bias, eps))
+}

candle-nn/src/lib.rs

@@ -15,9 +15,9 @@ pub mod vision;
 
 pub use activation::Activation;
 pub use conv::{Conv1d, Conv1dConfig};
-pub use embedding::Embedding;
+pub use embedding::{embedding, Embedding};
 pub use init::Init;
-pub use layer_norm::LayerNorm;
-pub use linear::Linear;
+pub use layer_norm::{layer_norm, LayerNorm};
+pub use linear::{linear, linear_no_bias, Linear};
 pub use optim::SGD;
-pub use var_builder::VarBuilder;
+pub use var_builder::{VarBuilder, VarMap};

candle-nn/src/linear.rs

@@ -17,7 +17,7 @@
 //! assert_eq!(ys.to_vec2::<f32>()?, &[[210.0, 430.0, 650.0]]);
 //! # Ok(()) }
 //! ```
-use candle::Tensor;
+use candle::{Result, Tensor};
 
 #[derive(Debug)]
 pub struct Linear {
@@ -42,3 +42,24 @@ impl Linear {
         }
     }
 }
+
+/// Create or initialize a new linear layer.
+///
+/// This uses some default names for weight and biases, namely `"weight"` and `"bias"`.
+pub fn linear(in_dim: usize, out_dim: usize, vs: crate::VarBuilder) -> Result<Linear> {
+    let init_ws = crate::init::DEFAULT_KAIMING_NORMAL;
+    let ws = vs.get_or_init((out_dim, in_dim), "weight", init_ws)?;
+    let bound = 1. / (in_dim as f64).sqrt();
+    let init_bs = crate::Init::Uniform {
+        lo: -bound,
+        up: bound,
+    };
+    let bs = vs.get_or_init(out_dim, "bias", init_bs)?;
+    Ok(Linear::new(ws, Some(bs)))
+}
+
+pub fn linear_no_bias(in_dim: usize, out_dim: usize, vs: crate::VarBuilder) -> Result<Linear> {
+    let init_ws = crate::init::DEFAULT_KAIMING_NORMAL;
+    let ws = vs.get_or_init((out_dim, in_dim), "weight", init_ws)?;
+    Ok(Linear::new(ws, None))
+}

candle-nn/src/optim.rs

@@ -8,7 +8,7 @@ pub struct SGD {
 }
 
 impl SGD {
-    pub fn new(vars: &[&Var], learning_rate: f64) -> Self {
+    pub fn from_slice(vars: &[&Var], learning_rate: f64) -> Self {
         let vars: Vec<_> = vars.iter().map(|&v| v.clone()).collect();
         Self {
             vars,
@@ -16,6 +16,13 @@ impl SGD {
         }
     }
 
+    pub fn new(vars: Vec<Var>, learning_rate: f64) -> Self {
+        Self {
+            vars,
+            learning_rate,
+        }
+    }
+
     pub fn empty(learning_rate: f64) -> Self {
         Self {
             vars: vec![],

candle-nn/src/var_builder.rs

@@ -1,7 +1,87 @@
-use candle::{safetensors::Load, DType, Device, Error, Result, Shape, Tensor};
+use candle::{safetensors::Load, DType, Device, Error, Result, Shape, Tensor, Var};
 use safetensors::{slice::IndexOp, tensor::SafeTensors};
 use std::collections::HashMap;
-use std::sync::Arc;
+use std::sync::{Arc, Mutex};
+
+/// A `VarMap` is a store that holds named variables. Variables can be retrieved from the stores
+/// and new variables can be added by providing some initialization config in case they are
+/// missing.
+/// `VarMap` structures can be serialized in the safetensors format.
+#[derive(Clone)]
+pub struct VarMap {
+    data: Arc<Mutex<HashMap<String, Var>>>,
+}
+
+impl VarMap {
+    /// Create a new empty `VarMap`.
+    #[allow(clippy::new_without_default)]
+    pub fn new() -> Self {
+        let data = Arc::new(Mutex::new(HashMap::new()));
+        Self { data }
+    }
+
+    /// Retrieve all the variables currently stored in the map.
+    pub fn all_vars(&self) -> Vec<Var> {
+        let tensor_data = self.data.lock().unwrap();
+        #[allow(clippy::map_clone)]
+        tensor_data.values().map(|c| c.clone()).collect::<Vec<_>>()
+    }
+
+    /// Save the map in the safetensors format.
+    pub fn save<P: AsRef<std::path::Path>>(&self, path: P) -> Result<()> {
+        let tensor_data = self.data.lock().unwrap();
+        let data = tensor_data.iter().map(|(k, v)| (k, v.as_tensor()));
+        safetensors::tensor::serialize_to_file(data, &None, path.as_ref())?;
+        Ok(())
+    }
+
+    /// Load some values from a safetensors file and modify the existing variables to have these
+    /// values.
+    ///
+    /// Note that values for variables that are currently not in the map are not kept.
+    pub fn load<P: AsRef<std::path::Path>>(&mut self, path: P) -> Result<()> {
+        let path = path.as_ref();
+        let data = unsafe { candle::safetensors::MmapedFile::new(path)? };
+        let data = data.deserialize()?;
+        let mut tensor_data = self.data.lock().unwrap();
+        for (name, var) in tensor_data.iter_mut() {
+            match data.tensor(name) {
+                Ok(data) => {
+                    let data: Tensor = data.load(var.device())?;
+                    if let Err(err) = var.set(&data) {
+                        candle::bail!("error setting {name} using data from {path:?}: {err}",)
+                    }
+                }
+                Err(_) => candle::bail!("cannot find tensor for {name}"),
+            }
+        }
+        Ok(())
+    }
+
+    /// Retrieve or add a new variable.
+    pub fn get<S: Into<Shape>>(
+        &self,
+        shape: S,
+        path: &str,
+        init: crate::Init,
+        dtype: DType,
+        device: &Device,
+    ) -> Result<Tensor> {
+        let shape = shape.into();
+        let mut tensor_data = self.data.lock().unwrap();
+        if let Some(tensor) = tensor_data.get(path) {
+            let tensor_shape = tensor.shape();
+            if &shape != tensor_shape {
+                candle::bail!("shape mismatch on {path}: {shape:?} <> {tensor_shape:?}")
+            }
+            return Ok(tensor.as_tensor().clone());
+        }
+        let var = init.var(shape, dtype, device)?;
+        let tensor = var.as_tensor().clone();
+        tensor_data.insert(path.to_string(), var);
+        Ok(tensor)
+    }
+}
 
 // TODO: Maybe we would want the storage to be generic, e.g. with Box<dyn> to avoid too many
 // generics.
@@ -13,6 +93,7 @@ enum Tensors<'a> {
     Npz(candle::npy::NpzTensors),
     TensorMap(HashMap<String, Tensor>),
     Zeros,
+    VarMap(VarMap),
 }
 
 struct TensorData<'a> {
@@ -64,6 +145,14 @@ impl<'a> TensorData<'a> {
             dtype,
         })
     }
+
+    fn from_varmap(varmap: &VarMap, dtype: DType, device: &Device) -> Self {
+        Self {
+            tensors: Tensors::VarMap(varmap.clone()),
+            device: device.clone(),
+            dtype,
+        }
+    }
 }
 
 #[derive(Clone)]
@@ -99,6 +188,14 @@ impl<'a> VarBuilder<'a> {
         }
     }
 
+    pub fn from_varmap(varmap: &VarMap, dtype: DType, device: &Device) -> Self {
+        let data = TensorData::from_varmap(varmap, dtype, device);
+        Self {
+            data: Arc::new(data),
+            path: vec![],
+        }
+    }
+
     pub fn from_npz<P: AsRef<std::path::Path>>(
         file: P,
         dtype: DType,
@@ -154,11 +251,7 @@ impl<'a> VarBuilder<'a> {
         world_size: usize,
     ) -> Result<Tensor> {
         let data = self.data.as_ref();
-        let path = if self.path.is_empty() {
-            tensor_name.to_string()
-        } else {
-            [&self.path.join("."), tensor_name].join(".")
-        };
+        let path = self.path(tensor_name);
         let tensor = match &self.data.tensors {
             Tensors::SafeTensorWithRouting {
                 routing,
@@ -205,19 +298,16 @@ impl<'a> VarBuilder<'a> {
                 let raw: Vec<u8> = iterator.into_iter().flatten().cloned().collect();
                 Tensor::from_raw_buffer(&raw, dtype, &shape, &data.device)?
             }
-            _ => unimplemented!(),
+            _ => candle::bail!("get_sharded is only available for safetensors"),
         };
         Ok(tensor)
     }
 
+    /// Retrieve the tensor associted with the current name and path.
     pub fn get<S: Into<Shape>>(&self, s: S, tensor_name: &str) -> Result<Tensor> {
         let data = self.data.as_ref();
         let s: Shape = s.into();
-        let path = if self.path.is_empty() {
-            tensor_name.to_string()
-        } else {
-            [&self.path.join("."), tensor_name].join(".")
-        };
+        let path = self.path(tensor_name);
         let tensor = match &self.data.tensors {
             Tensors::Zeros => Tensor::zeros(&s, data.dtype, &data.device)?.contiguous()?,
             Tensors::TensorMap(ts) => ts
@@ -229,6 +319,18 @@ impl<'a> VarBuilder<'a> {
                     .bt()
                 })?
                 .clone(),
+            Tensors::VarMap(varmap) => {
+                let data = varmap.data.lock().unwrap();
+                data.get(&path)
+                    .ok_or_else(|| {
+                        Error::CannotFindTensor {
+                            path: path.to_string(),
+                        }
+                        .bt()
+                    })?
+                    .as_tensor()
+                    .clone()
+            }
             Tensors::Npz(npz) => npz.get(&path)?.ok_or_else(|| {
                 Error::CannotFindTensor {
                     path: path.to_string(),
@@ -261,4 +363,32 @@ impl<'a> VarBuilder<'a> {
         }
         Ok(tensor)
     }
+
+    /// Retrieve the tensor associted with the current name and path or initialize a new tensor if
+    /// it's missing.
+    ///
+    /// Tensor initialization is only available if the `VarBuilder` is backed by a `VarMap`.
+    pub fn get_or_init<S: Into<Shape>>(
+        &self,
+        s: S,
+        tensor_name: &str,
+        init: crate::Init,
+    ) -> Result<Tensor> {
+        let data = self.data.as_ref();
+        match &self.data.tensors {
+            Tensors::VarMap(varmap) => {
+                let path = self.path(tensor_name);
+                varmap.get(s, &path, init, data.dtype, &data.device)
+            }
+            _ => self.get(s, tensor_name),
+        }
+    }
+
+    fn path(&self, tensor_name: &str) -> String {
+        if self.path.is_empty() {
+            tensor_name.to_string()
+        } else {
+            [&self.path.join("."), tensor_name].join(".")
+        }
+    }
 }

candle-nn/tests/optim.rs

@@ -8,7 +8,7 @@ use candle_nn::{Linear, SGD};
 #[test]
 fn sgd_optim() -> Result<()> {
     let x = Var::new(0f32, &Device::Cpu)?;
-    let sgd = SGD::new(&[&x], 0.1);
+    let sgd = SGD::new(vec![x.clone()], 0.1);
     let xt = x.as_tensor();
     for _step in 0..100 {
         let loss = ((xt - 4.2)? * (xt - 4.2)?)?;
@@ -54,7 +54,7 @@ fn sgd_linear_regression() -> Result<()> {
     // Now use backprop to run a linear regression between samples and get the coefficients back.
     let w = Var::new(&[[0f32, 0.]], &Device::Cpu)?;
     let b = Var::new(0f32, &Device::Cpu)?;
-    let sgd = SGD::new(&[&w, &b], 0.004);
+    let sgd = SGD::new(vec![w.clone(), b.clone()], 0.004);
     let lin = Linear::new(w.as_tensor().clone(), Some(b.as_tensor().clone()));
     for _step in 0..1000 {
         let ys = lin.forward(&sample_xs)?;