Add a KV cache to whisper. (#426)

candle-examples/examples/whisper/main.rs

@@ -109,8 +109,8 @@ impl Decoder {
     }
 
     fn decode(&mut self, mel: &Tensor, t: f64) -> Result<DecodingResult> {
-        let model = &self.model;
-        let audio_features = model.encoder.forward(mel)?;
+        let model = &mut self.model;
+        let audio_features = model.encoder.forward(mel, true)?;
         println!("audio features: {:?}", audio_features.dims());
         let sample_len = model.config.max_target_positions / 2;
         let mut sum_logprob = 0f64;
@@ -126,7 +126,7 @@ impl Decoder {
             // The model expects a batch dim but this inference loop does not handle
             // it so we add it at this point.
             let tokens_t = tokens_t.unsqueeze(0)?;
-            let logits = model.decoder.forward(&tokens_t, &audio_features)?;
+            let logits = model.decoder.forward(&tokens_t, &audio_features, i == 0)?;
             let logits = logits.squeeze(0)?;
 
             // Extract the no speech probability on the first iteration by looking at the first
@@ -393,10 +393,10 @@ fn main() -> Result<()> {
     let weights = weights.deserialize()?;
     let vb = VarBuilder::from_safetensors(vec![weights], DTYPE, &device);
     let config: Config = serde_json::from_str(&std::fs::read_to_string(config_filename)?)?;
-    let model = Whisper::load(&vb, config)?;
+    let mut model = Whisper::load(&vb, config)?;
 
     let language_token = match (args.model.is_multilingual(), args.language) {
-        (true, None) => Some(multilingual::detect_language(&model, &tokenizer, &mel)?),
+        (true, None) => Some(multilingual::detect_language(&mut model, &tokenizer, &mel)?),
         (false, None) => None,
         (true, Some(language)) => match token_id(&tokenizer, &format!("<|{language}|>")) {
             Ok(token_id) => Some(token_id),

candle-examples/examples/whisper/model.rs

@@ -105,6 +105,7 @@ struct MultiHeadAttention {
     out: Linear,
     n_head: usize,
     span: tracing::Span,
+    kv_cache: Option<(Tensor, Tensor)>,
 }
 
 impl MultiHeadAttention {
@@ -121,14 +122,39 @@ impl MultiHeadAttention {
             out,
             n_head,
             span,
+            kv_cache: None,
         })
     }
 
-    fn forward(&self, x: &Tensor, xa: Option<&Tensor>, mask: Option<&Tensor>) -> Result<Tensor> {
+    fn forward(
+        &mut self,
+        x: &Tensor,
+        xa: Option<&Tensor>,
+        mask: Option<&Tensor>,
+        flush_cache: bool,
+    ) -> Result<Tensor> {
         let _enter = self.span.enter();
         let q = self.query.forward(x)?;
-        let k = self.key.forward(xa.unwrap_or(x))?;
-        let v = self.value.forward(xa.unwrap_or(x))?;
+        let (k, v) = match xa {
+            None => {
+                let k = self.key.forward(x)?;
+                let v = self.value.forward(x)?;
+                (k, v)
+            }
+            Some(x) => {
+                if flush_cache {
+                    self.kv_cache = None;
+                }
+                if let Some((k, v)) = &self.kv_cache {
+                    (k.clone(), v.clone())
+                } else {
+                    let k = self.key.forward(x)?;
+                    let v = self.value.forward(x)?;
+                    self.kv_cache = Some((k.clone(), v.clone()));
+                    (k, v)
+                }
+            }
+        };
         let wv = self.qkv_attention(&q, &k, &v, mask)?;
         let out = self.out.forward(&wv)?;
         Ok(out)
@@ -201,12 +227,20 @@ impl ResidualAttentionBlock {
         })
     }
 
-    fn forward(&self, x: &Tensor, xa: Option<&Tensor>, mask: Option<&Tensor>) -> Result<Tensor> {
+    fn forward(
+        &mut self,
+        x: &Tensor,
+        xa: Option<&Tensor>,
+        mask: Option<&Tensor>,
+        flush_kv_cache: bool,
+    ) -> Result<Tensor> {
         let _enter = self.span.enter();
-        let attn = self.attn.forward(&self.attn_ln.forward(x)?, None, mask)?;
+        let attn = self
+            .attn
+            .forward(&self.attn_ln.forward(x)?, None, mask, flush_kv_cache)?;
         let mut x = (x + attn)?;
-        if let Some((attn, ln)) = &self.cross_attn {
-            x = (&x + attn.forward(&ln.forward(&x)?, xa, None)?)?;
+        if let Some((attn, ln)) = &mut self.cross_attn {
+            x = (&x + attn.forward(&ln.forward(&x)?, xa, None, flush_kv_cache)?)?;
         }
         let mlp = self.mlp_linear2.forward(
             &self
@@ -283,7 +317,7 @@ impl AudioEncoder {
         })
     }
 
-    pub fn forward(&self, x: &Tensor) -> Result<Tensor> {
+    pub fn forward(&mut self, x: &Tensor, flush_kv_cache: bool) -> Result<Tensor> {
         let _enter = self.span.enter();
         let x = {
             let _enter = self.conv1_span.enter();
@@ -297,8 +331,8 @@ impl AudioEncoder {
         let (_bsize, seq_len, _hidden) = x.dims3()?;
         let positional_embedding = self.positional_embedding.narrow(0, 0, seq_len)?;
         let mut x = x.broadcast_add(&positional_embedding)?;
-        for block in self.blocks.iter() {
-            x = block.forward(&x, None, None)?
+        for block in self.blocks.iter_mut() {
+            x = block.forward(&x, None, None, flush_kv_cache)?
         }
         let x = self.ln_post.forward(&x)?;
         Ok(x)
@@ -344,15 +378,15 @@ impl TextDecoder {
         })
     }
 
-    pub fn forward(&self, x: &Tensor, xa: &Tensor) -> Result<Tensor> {
+    pub fn forward(&mut self, x: &Tensor, xa: &Tensor, flush_kv_cache: bool) -> Result<Tensor> {
         let _enter = self.span.enter();
         let x_dims = x.dims();
         let last = x_dims[x_dims.len() - 1];
         let token_embedding = self.token_embedding.forward(x)?;
         let positional_embedding = self.positional_embedding.narrow(0, 0, last)?;
         let mut x = token_embedding.broadcast_add(&positional_embedding)?;
-        for block in self.blocks.iter() {
-            x = block.forward(&x, Some(xa), Some(&self.mask))?;
+        for block in self.blocks.iter_mut() {
+            x = block.forward(&x, Some(xa), Some(&self.mask), flush_kv_cache)?;
         }
         let x = self.ln.forward(&x)?;
         let w = self
@@ -383,9 +417,14 @@ impl Whisper {
     }
 
     #[allow(dead_code)]
-    pub fn forward(&self, mel: &Tensor, tokens: &Tensor) -> Result<Tensor> {
-        let enc = self.encoder.forward(mel)?;
-        let dec = self.decoder.forward(tokens, &enc)?;
+    pub fn forward(
+        &mut self,
+        mel: &Tensor,
+        tokens: &Tensor,
+        flush_kv_cache: bool,
+    ) -> Result<Tensor> {
+        let enc = self.encoder.forward(mel, flush_kv_cache)?;
+        let dec = self.decoder.forward(tokens, &enc, flush_kv_cache)?;
         Ok(dec)
     }
 }

candle-examples/examples/whisper/multilingual.rs

@@ -105,19 +105,19 @@ const LANGUAGES: [(&str, &str); 99] = [
 ];
 
 /// Returns the token id for the selected language.
-pub fn detect_language(model: &Whisper, tokenizer: &Tokenizer, mel: &Tensor) -> Result<u32> {
+pub fn detect_language(model: &mut Whisper, tokenizer: &Tokenizer, mel: &Tensor) -> Result<u32> {
     let device = mel.device();
     let language_token_ids = LANGUAGES
         .iter()
         .map(|(t, _)| crate::token_id(tokenizer, &format!("<|{t}|>")))
         .collect::<Result<Vec<_>>>()?;
     let sot_token = crate::token_id(tokenizer, crate::SOT_TOKEN)?;
-    let audio_features = model.encoder.forward(mel)?;
+    let audio_features = model.encoder.forward(mel, true)?;
     let tokens = Tensor::new(&[[sot_token]], device)?;
     let language_token_ids = Tensor::new(language_token_ids.as_slice(), device)?;
     let logits = model
         .decoder
-        .forward(&tokens, &audio_features)?
+        .forward(&tokens, &audio_features, true)?
         .i(0)?
         .i(0)?;
     let logits = logits.index_select(&language_token_ids, 0)?;