From 7e2caab17178f967c550f76a53b86e09707dc41d Mon Sep 17 00:00:00 2001 From: Daniel Han Date: Wed, 12 Feb 2025 19:11:09 -0800 Subject: [PATCH] Update llama.py --- unsloth/models/llama.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/unsloth/models/llama.py b/unsloth/models/llama.py index 452bb78e27..e04c573c6c 100644 --- a/unsloth/models/llama.py +++ b/unsloth/models/llama.py @@ -449,7 +449,7 @@ def LlamaAttention_fast_forward( else: # Grouped query attention if SDPA_HAS_GQA: - print(attention_mask.shape, Q.shape, K.shape, V.shape) + print(attention_mask.shape, Q.shape, K.shape, V.shape, attention_mask.dtype) # Needs (batch_size, n_heads, seq_len, head_dim) # is_casual and attention_mask must not be both set! A = scaled_dot_product_attention(Q, K, V, attn_mask = attention_mask, is_causal = False, enable_gqa = n_groups != 1)