Debugging

mistralrs-core/src/vision_models/mllama/text.rs

@@ -258,6 +258,7 @@ impl MLlamaSelfAttentionDecoderLayer {
  if hidden_states.dtype() == DType::F16 {
  hidden_states = clamp_inf(&hidden_states)?;
  }
+ dbg!("self attn",&hidden_states.mean_all());
 
  let residual = &hidden_states;
  let mut hidden_states = self.post_attention_layernorm.forward(&hidden_states)?;
@@ -266,6 +267,7 @@ impl MLlamaSelfAttentionDecoderLayer {
  if hidden_states.dtype() == DType::F16 {
  hidden_states = clamp_inf(&hidden_states)?;
  }
+ dbg!("self mlp",&hidden_states.mean_all());
 
  residual + hidden_states
  }
@@ -495,6 +497,7 @@ impl MLlamaCrossAttentionDecoderLayer {
  if hidden_states.dtype() == DType::F16 {
  hidden_states = clamp_inf(&hidden_states)?;
  }
+ dbg!("cross attn",&hidden_states.mean_all());
 
  let residual = &hidden_states;
  let mut hidden_states = self.post_attention_layernorm.forward(&hidden_states)?;
@@ -503,12 +506,14 @@ impl MLlamaCrossAttentionDecoderLayer {
  if hidden_states.dtype() == DType::F16 {
  hidden_states = clamp_inf(&hidden_states)?;
  }
+ dbg!("cross mlp",&hidden_states.mean_all());
  if let Some(full_text_row_masked_out_mask) = full_text_row_masked_out_mask {
  hidden_states = full_text_row_masked_out_mask
  .to_dtype(hidden_states.dtype())?
  .i((.., 0))?
  .broadcast_mul(&hidden_states)?;
  }
+ dbg!("ftrm",&hidden_states.mean_all());
 
  residual + hidden_states.broadcast_mul(&self.mlp_gate)?
  }
@@ -691,7 +696,7 @@ impl MLlamaTextModel {
  )?;
  }
  }
- dbg!(&hidden_states);
+ dbg!("eol",&hidden_states.mean_all());
  }
 
  hidden_states = hidden_states.to_device(&self.device)?;