Flan T5: Read lm_head when word embeddings are not tied (#903)

* Read lm_head when word embeddings are not tied

* Fix formatting

* Address comments

candle-transformers/src/models/t5.rs

@@ -18,12 +18,15 @@ fn default_use_cache() -> bool {
     true
 }
 
+fn default_tie_word_embeddings() -> bool {
+    true
+}
+
 fn get_mask(size: usize, device: &Device) -> Result<Tensor> {
     let mask: Vec<_> = (0..size)
         .flat_map(|i| (0..size).map(move |j| u8::from(j > i)))
         .collect();
-    let result = Tensor::from_slice(&mask, (size, size), device)?;
-    Ok(result)
+    Tensor::from_slice(&mask, (size, size), device)
 }
 
 fn masked_fill(on_false: &Tensor, mask: &Tensor, on_true: f32) -> Result<Tensor> {
@@ -50,6 +53,8 @@ pub struct Config {
     initializer_factor: f64,
     #[serde(default)]
     feed_forward_proj: Activation,
+    #[serde(default = "default_tie_word_embeddings")]
+    tie_word_embeddings: bool,
     #[serde(default = "default_is_decoder")]
     is_decoder: bool,
     is_encoder_decoder: bool,
@@ -75,6 +80,7 @@ impl Default for Config {
             layer_norm_epsilon: 1e-6,
             initializer_factor: 1.0,
             feed_forward_proj: Activation::Relu,
+            tie_word_embeddings: true,
             is_decoder: false,
             is_encoder_decoder: true,
             use_cache: true,
@@ -94,6 +100,7 @@ impl Config {
             dropout_rate: 0.1,
             eos_token_id: 1,
             feed_forward_proj: Activation::Relu,
+            tie_word_embeddings: true,
             initializer_factor: 1.0,
             is_decoder: false,
             is_encoder_decoder: true,
@@ -611,6 +618,9 @@ impl T5EncoderModel {
 pub struct T5ForConditionalGeneration {
     encoder: T5Stack,
     decoder: T5Stack,
+    d_model: usize,
+    tie_word_embeddings: bool,
+    lm_head: Option<Linear>,
     shared: Arc<Embedding>,
     device: Device,
 }
@@ -618,6 +628,7 @@ pub struct T5ForConditionalGeneration {
 impl T5ForConditionalGeneration {
     pub fn load(vb: VarBuilder, cfg: &Config) -> Result<Self> {
         assert!(cfg.is_encoder_decoder);
+        let d_model = cfg.d_model;
         let shared = embedding(cfg.vocab_size, cfg.d_model, vb.pp("shared"))?;
         let shared = Arc::new(shared);
 
@@ -633,9 +644,23 @@ impl T5ForConditionalGeneration {
         decoder_cfg.num_layers = cfg.num_decoder_layers.unwrap_or(cfg.num_layers);
         let decoder = T5Stack::load(true, vb.pp("decoder"), &shared, &decoder_cfg)?;
 
+        let tie_word_embeddings = cfg.tie_word_embeddings;
+        let lm_head = if tie_word_embeddings {
+            None
+        } else {
+            Some(linear_no_bias(
+                cfg.d_model,
+                cfg.vocab_size,
+                vb.pp("lm_head"),
+            )?)
+        };
+
         Ok(Self {
             encoder,
             decoder,
+            d_model,
+            tie_word_embeddings,
+            lm_head,
             shared,
             device: vb.device().clone(),
         })
@@ -653,12 +678,23 @@ impl T5ForConditionalGeneration {
         let decoder_output = self
             .decoder
             .forward(decoder_input_ids, Some(encoder_output))?;
-        let sequence_output = decoder_output
+
+        let scaling_factor = if self.tie_word_embeddings {
+            // Rescale output before projecting on vocab
+            // See https://github.com/tensorflow/mesh/blob/fa19d69eafc9a482aff0b59ddd96b025c0cb207d/mesh_tensorflow/transformer/transformer.py#L586
+            (self.d_model as f64).sqrt()
+        } else {
+            1.0
+        };
+        let sequence_output = ((decoder_output
             .narrow(1, decoder_output.dim(1)? - 1, 1)?
-            .squeeze(1)?;
-        // TODO: check cfg.tie_word_embeddings to load from model instead.
-        let lm_head_weights = self.shared.embeddings().t()?;
-        let output = sequence_output.matmul(&lm_head_weights)?;
+            .squeeze(1)?)
+            * scaling_factor)?;
+        let output = match self.lm_head {
+            None => sequence_output.matmul(&self.shared.embeddings().t()?)?,
+            Some(ref lm_head) => lm_head.forward(&sequence_output)?,
+        };
+
         // TODO: Rescale output before projecting on vocab? * (self.model_dim**-0.5)
         Ok(output)
     }