Add some optional repeat penalty. (#623)

* Add some optional repeat penalty.

* Add the missing files.

candle-examples/examples/llama/main.rs

@@ -83,6 +83,14 @@ struct Args {
     /// (same structure as huggingface online)
     #[arg(long)]
     local_weights: Option<String>,
+
+    /// Penalty to be applied for repeating tokens, 1. means no penalty.
+    #[arg(long, default_value_t = 1.0)]
+    repeat_penalty: f32,
+
+    /// The context size to consider for the repeat penalty.
+    #[arg(long, default_value_t = 64)]
+    repeat_last_n: usize,
 }
 
 fn main() -> Result<()> {
@@ -200,6 +208,16 @@ fn main() -> Result<()> {
         let input = Tensor::new(ctxt, &device)?.unsqueeze(0)?;
         let logits = llama.forward(&input, index_pos)?;
         let logits = logits.squeeze(0)?;
+        let logits = if args.repeat_penalty == 1. {
+            logits
+        } else {
+            let start_at = tokens.len().saturating_sub(args.repeat_last_n);
+            candle_transformers::utils::apply_repeat_penalty(
+                &logits,
+                args.repeat_penalty,
+                &tokens[start_at..],
+            )?
+        };
         index_pos += ctxt.len();
 
         let next_token = logits_processor.sample(&logits)?;

candle-examples/examples/quantized/main.rs

@@ -533,22 +533,6 @@ fn print_token(next_token: u32, tokenizer: &Tokenizer) {
     }
 }
 
-fn apply_repeat_penalty(logits: &Tensor, penalty: f32, context: &[u32]) -> Result<Tensor> {
-    let mut logits = logits.to_vec1::<f32>()?;
-    let context: std::collections::HashSet<_> = context.iter().collect();
-    for (token_id, logit) in logits.iter_mut().enumerate() {
-        if context.contains(&(token_id as u32)) {
-            if *logit >= 0. {
-                *logit /= penalty
-            } else {
-                *logit *= penalty
-            }
-        }
-    }
-    let logits_len = logits.len();
-    Tensor::from_vec(logits, logits_len, &Device::Cpu)
-}
-
 fn format_size(size_in_bytes: usize) -> String {
     if size_in_bytes < 1_000 {
         format!("{}B", size_in_bytes)
@@ -670,7 +654,11 @@ fn main() -> anyhow::Result<()> {
             logits
         } else {
             let start_at = all_tokens.len().saturating_sub(args.repeat_last_n);
-            apply_repeat_penalty(&logits, args.repeat_penalty, &all_tokens[start_at..])?
+            candle_transformers::utils::apply_repeat_penalty(
+                &logits,
+                args.repeat_penalty,
+                &all_tokens[start_at..],
+            )?
         };
         next_token = logits_processor.sample(&logits)?;
         all_tokens.push(next_token);

candle-transformers/src/lib.rs

@@ -1,3 +1,4 @@
 pub mod generation;
 pub mod models;
 pub mod pipelines;
+pub mod utils;

candle-transformers/src/utils.rs

@@ -0,0 +1,18 @@
+use candle::{Result, Tensor};
+
+pub fn apply_repeat_penalty(logits: &Tensor, penalty: f32, context: &[u32]) -> Result<Tensor> {
+    let device = logits.device();
+    let mut logits = logits.to_vec1::<f32>()?;
+    let context: std::collections::HashSet<_> = context.iter().collect();
+    for (token_id, logit) in logits.iter_mut().enumerate() {
+        if context.contains(&(token_id as u32)) {
+            if *logit >= 0. {
+                *logit /= penalty
+            } else {
+                *logit *= penalty
+            }
+        }
+    }
+    let logits_len = logits.len();
+    Tensor::from_vec(logits, logits_len, device)
+}