model: Pass input tensor instead of raw data to models

Rather than directly giving the input data to models, we can
pass a tensor instead. In the short term, this saves some duplicated
code.

Longer term, we will want to overlap setting up the next batch with
processing of the current one. In this case, we will only have the
shape of tensor but it will not be loaded with data at the time of
graph generation. By passing only a tensor to models now, we set up
this possibility and prevent them from relying on data that they won't
have in the future.

Although the same could be done for Positions and Outputs, in some
cases we either need the raw input data or don't use them at all.
Therefore, for now we leave them as they are and allow models to
convert them to tensors as needed.

model/input/input.go

@@ -1,5 +1,7 @@
 package input
 
+import "github.com/ollama/ollama/ml"
+
 // Input represents one token in the input stream
 type Input struct {
 	// Token is a single element of text.
@@ -36,7 +38,7 @@ type MultimodalIndex struct {
 // Batch contains the inputs for a model forward pass
 type Batch struct {
 	// Inputs is the input tokens, including placeholders for multimodal inputs.
-	Inputs []int32
+	Inputs ml.Tensor
 
 	// Multimodal is a set of multimodal embeddings previously created by
 	// EncodeMultimodal, along with an index into Inputs. Unused for text-only