Remove some unnecessary &muts (#473)

* Remove some &muts from the cache traits * Remove from model forward methods
EricLBuehler · Jun 24, 2024 · a1e41aa · a1e41aa
1 parent 3febef2
commit a1e41aa
Show file tree

Hide file tree

Showing 29 changed files with 156 additions and 164 deletions.
diff --git a/mistralrs-core/src/models/gemma.rs b/mistralrs-core/src/models/gemma.rs
@@ -161,7 +161,7 @@ impl Attention {
     }
 
     fn forward(
-        &mut self,
+        &self,
         xs: &Tensor,
         attention_mask: Option<&Tensor>,
         seqlen_offsets: &[usize],
@@ -277,7 +277,7 @@ impl DecoderLayer {
     }
 
     fn forward(
-        &mut self,
+        &self,
         xs: &Tensor,
         attention_mask: Option<&Tensor>,
         seqlen_offsets: &[usize],
@@ -374,7 +374,7 @@ impl Model {
     }
 
     pub fn forward(
-        &mut self,
+        &self,
         input_ids: &Tensor,
         seqlen_offsets: &[usize],
         start_offsets_kernel: Tensor,
@@ -389,7 +389,7 @@ impl Model {
             xs.dtype(),
             self.layers[0].self_attn.num_heads,
         )?;
-        for (i, layer) in self.layers.iter_mut().enumerate() {
+        for (i, layer) in self.layers.iter().enumerate() {
             xs = self.mapper.map(xs, i)?;
             xs = layer.forward(
                 &xs,
@@ -430,7 +430,7 @@ impl IsqModel for Model {
 
 impl NormalModel for Model {
     fn forward(
-        &mut self,
+        &self,
         input_ids: &Tensor,
         seqlen_offsets: &[usize],
         start_offsets_kernel: Tensor,
@@ -445,7 +445,7 @@ impl NormalModel for Model {
         )
     }
     fn xlora_forward(
-        &mut self,
+        &self,
         _input_ids: &Tensor,
         _input_ids_full: &Tensor,
         _seqlen_offsets: &[usize],

diff --git a/mistralrs-core/src/models/llama.rs b/mistralrs-core/src/models/llama.rs
@@ -255,7 +255,7 @@ pub struct Llama {
 
 impl Llama {
     pub fn forward(
-        &mut self,
+        &self,
         input_ids: &Tensor,
         seqlen_offsets: &[usize],
         start_offsets_kernel: Tensor,
@@ -374,7 +374,7 @@ impl IsqModel for Llama {
 
 impl NormalModel for Llama {
     fn forward(
-        &mut self,
+        &self,
         input_ids: &Tensor,
         seqlen_offsets: &[usize],
         start_offsets_kernel: Tensor,
@@ -389,7 +389,7 @@ impl NormalModel for Llama {
         )
     }
     fn xlora_forward(
-        &mut self,
+        &self,
         _input_ids: &Tensor,
         _input_ids_full: &Tensor,
         _seqlen_offsets: &[usize],

diff --git a/mistralrs-core/src/models/mistral.rs b/mistralrs-core/src/models/mistral.rs
@@ -113,7 +113,7 @@ impl Attention {
     }
 
     fn forward(
-        &mut self,
+        &self,
         xs: &Tensor,
         attention_mask: Option<&Tensor>,
         seqlen_offsets: &[usize],
@@ -238,7 +238,7 @@ impl DecoderLayer {
     }
 
     fn forward(
-        &mut self,
+        &self,
         xs: &Tensor,
         attention_mask: Option<&Tensor>,
         seqlen_offsets: &[usize],
@@ -356,7 +356,7 @@ impl Model {
     }
 
     pub fn forward(
-        &mut self,
+        &self,
         input_ids: &Tensor,
         seqlen_offsets: &[usize],
         start_offsets_kernel: Tensor,
@@ -372,7 +372,7 @@ impl Model {
     }
 
     pub fn forward_embeds(
-        &mut self,
+        &self,
         input_ids: &Tensor,
         input_embeds: Tensor,
         seqlen_offsets: &[usize],
@@ -388,7 +388,7 @@ impl Model {
             xs.dtype(),
             self.layers[0].self_attn.num_heads,
         )?;
-        for (i, layer) in self.layers.iter_mut().enumerate() {
+        for (i, layer) in self.layers.iter().enumerate() {
             xs = self.mapper.map(xs, i)?;
             xs = layer.forward(
                 &xs,
@@ -429,7 +429,7 @@ impl IsqModel for Model {
 
 impl NormalModel for Model {
     fn forward(
-        &mut self,
+        &self,
         input_ids: &Tensor,
         seqlen_offsets: &[usize],
         start_offsets_kernel: Tensor,
@@ -444,7 +444,7 @@ impl NormalModel for Model {
         )
     }
     fn xlora_forward(
-        &mut self,
+        &self,
         _input_ids: &Tensor,
         _input_ids_full: &Tensor,
         _seqlen_offsets: &[usize],

diff --git a/mistralrs-core/src/models/mixtral.rs b/mistralrs-core/src/models/mixtral.rs
@@ -77,7 +77,7 @@ impl Attention {
     }
 
     fn forward(
-        &mut self,
+        &self,
         xs: &Tensor,
         attention_mask: Option<&Tensor>,
         seqlen_offsets: &[usize],
@@ -339,7 +339,7 @@ impl DecoderLayer {
     }
 
     fn forward(
-        &mut self,
+        &self,
         xs: &Tensor,
         attention_mask: Option<&Tensor>,
         seqlen_offsets: &[usize],
@@ -443,7 +443,7 @@ impl Model {
     }
 
     pub fn forward(
-        &mut self,
+        &self,
         input_ids: &Tensor,
         seqlen_offsets: &[usize],
         start_offsets_kernel: Tensor,
@@ -458,7 +458,7 @@ impl Model {
             xs.dtype(),
             self.layers[0].self_attn.num_heads,
         )?;
-        for (i, layer) in self.layers.iter_mut().enumerate() {
+        for (i, layer) in self.layers.iter().enumerate() {
             xs = self.mapper.map(xs, i)?;
             xs = layer.forward(
                 &xs,
@@ -502,7 +502,7 @@ impl IsqModel for Model {
 
 impl NormalModel for Model {
     fn forward(
-        &mut self,
+        &self,
         input_ids: &Tensor,
         seqlen_offsets: &[usize],
         start_offsets_kernel: Tensor,
@@ -517,7 +517,7 @@ impl NormalModel for Model {
         )
     }
     fn xlora_forward(
-        &mut self,
+        &self,
         _input_ids: &Tensor,
         _input_ids_full: &Tensor,
         _seqlen_offsets: &[usize],

diff --git a/mistralrs-core/src/models/phi2.rs b/mistralrs-core/src/models/phi2.rs
@@ -130,7 +130,7 @@ impl Attention {
     }
 
     fn forward(
-        &mut self,
+        &self,
         xs: &Tensor,
         mask: Option<&Tensor>,
         seqlen_offsets: &[usize],
@@ -253,7 +253,7 @@ impl DecoderLayer {
     }
 
     fn forward(
-        &mut self,
+        &self,
         xs: &Tensor,
         mask: Option<&Tensor>,
         seqlen_offsets: &[usize],
@@ -347,7 +347,7 @@ impl Model {
     }
 
     pub fn forward(
-        &mut self,
+        &self,
         input_ids: &Tensor,
         seqlen_offsets: &[usize],
         start_offsets_kernel: Tensor,
@@ -361,7 +361,7 @@ impl Model {
             xs.dtype(),
             self.layers[0].self_attn.num_heads,
         )?;
-        for (i, layer) in self.layers.iter_mut().enumerate() {
+        for (i, layer) in self.layers.iter().enumerate() {
             xs = self.mapper.map(xs, i)?;
             xs = layer.forward(
                 &xs,
@@ -400,7 +400,7 @@ impl IsqModel for Model {
 
 impl NormalModel for Model {
     fn forward(
-        &mut self,
+        &self,
         input_ids: &Tensor,
         seqlen_offsets: &[usize],
         start_offsets_kernel: Tensor,
@@ -415,7 +415,7 @@ impl NormalModel for Model {
         )
     }
     fn xlora_forward(
-        &mut self,
+        &self,
         _input_ids: &Tensor,
         _input_ids_full: &Tensor,
         _seqlen_offsets: &[usize],

diff --git a/mistralrs-core/src/models/phi3.rs b/mistralrs-core/src/models/phi3.rs
@@ -91,7 +91,7 @@ impl Attention {
     }
 
     fn forward(
-        &mut self,
+        &self,
         xs: &Tensor,
         attention_mask: Option<&Tensor>,
         seqlen_offsets: &[usize],
@@ -254,7 +254,7 @@ impl DecoderLayer {
     }
 
     fn forward(
-        &mut self,
+        &self,
         xs: &Tensor,
         attention_mask: Option<&Tensor>,
         seqlen_offsets: &[usize],
@@ -348,7 +348,7 @@ impl Model {
     }
 
     pub fn forward(
-        &mut self,
+        &self,
         input_ids: &Tensor,
         seqlen_offsets: &[usize],
         position_ids: &[usize],
@@ -364,7 +364,7 @@ impl Model {
             self.layers[0].self_attn.num_heads,
         )?;
 
-        for (i, layer) in self.layers.iter_mut().enumerate() {
+        for (i, layer) in self.layers.iter().enumerate() {
             xs = self.mapper.map(xs, i)?;
             xs = layer.forward(
                 &xs,
@@ -402,7 +402,7 @@ impl IsqModel for Model {
 
 impl NormalModel for Model {
     fn forward(
-        &mut self,
+        &self,
         input_ids: &Tensor,
         seqlen_offsets: &[usize],
         _start_offsets_kernel: Tensor,
@@ -412,7 +412,7 @@ impl NormalModel for Model {
         self.forward(input_ids, seqlen_offsets, &position_ids, context_lens)
     }
     fn xlora_forward(
-        &mut self,
+        &self,
         _input_ids: &Tensor,
         _input_ids_full: &Tensor,
         _seqlen_offsets: &[usize],

diff --git a/mistralrs-core/src/models/quantized_llama.rs b/mistralrs-core/src/models/quantized_llama.rs
@@ -127,7 +127,7 @@ struct LayerWeights {
 
 impl LayerWeights {
     fn forward_attn(
-        &mut self,
+        &self,
         x: &Tensor,
         mask: Option<&Tensor>,
         start_offsets: &[usize],
@@ -484,7 +484,7 @@ impl ModelConfig::FromGGUF for ModelWeights {
 
 impl ModelWeights {
     pub fn forward(
-        &mut self,
+        &self,
         x: &Tensor,
         start_offsets: &[usize],
         start_offsets_kernel: Tensor,
@@ -498,7 +498,7 @@ impl ModelWeights {
             DType::F32,
             self.layers[0].n_head,
         )?;
-        for (i, layer) in self.layers.iter_mut().enumerate() {
+        for (i, layer) in self.layers.iter().enumerate() {
             if let Some(ref mapper) = self.mapper {
                 layer_in = mapper.map(layer_in, i)?;
             }

diff --git a/mistralrs-core/src/models/quantized_phi2.rs b/mistralrs-core/src/models/quantized_phi2.rs
@@ -58,7 +58,7 @@ impl LayerWeights {
     }
 
     fn forward_attn(
-        &mut self,
+        &self,
         x: &Tensor,
         mask: Option<&Tensor>,
         seqlen_offsets: &[usize],
@@ -268,7 +268,7 @@ impl ModelConfig::FromGGUF for ModelWeights {
 
 impl ModelWeights {
     pub fn forward(
-        &mut self,
+        &self,
         input_ids: &Tensor,
         seqlen_offsets: &[usize],
         context_lens: Vec<(usize, usize)>,
@@ -281,7 +281,7 @@ impl ModelWeights {
             DType::F32,
             self.layers[0].n_head,
         )?;
-        for (i, layer) in self.layers.iter_mut().enumerate() {
+        for (i, layer) in self.layers.iter().enumerate() {
             xs = self.mapper.map(xs, i)?;
             let residual = &xs;
             let xs_norm = xs.apply(&layer.attn_norm)?;

diff --git a/mistralrs-core/src/models/quantized_phi3.rs b/mistralrs-core/src/models/quantized_phi3.rs
@@ -67,7 +67,7 @@ impl LayerWeights {
     }
 
     fn forward_attn(
-        &mut self,
+        &self,
         x: &Tensor,
         mask: Option<&Tensor>,
         seqlen_offsets: &[usize],
@@ -303,7 +303,7 @@ impl ModelConfig::FromGGUF for ModelWeights {
 }
 
 impl ModelWeights {
-    pub fn forward(&mut self, input_ids: &Tensor, seqlen_offsets: &[usize]) -> Result<Tensor> {
+    pub fn forward(&self, input_ids: &Tensor, seqlen_offsets: &[usize]) -> Result<Tensor> {
         let (_b_sz, seq_len) = input_ids.dims2()?;
         let mut xs = self.tok_embeddings.forward(input_ids)?;
         let mut cache = self.cache.lock();
@@ -314,7 +314,7 @@ impl ModelWeights {
             DType::F32,
             self.layers[0].n_head,
         )?;
-        for (i, layer) in self.layers.iter_mut().enumerate() {
+        for (i, layer) in self.layers.iter().enumerate() {
             if let Some(ref mapper) = self.mapper {
                 xs = mapper.map(xs, i)?;
             }