Merge pull request #3592 from AI-Hypercomputer:agagik-gemma4-flops

PiperOrigin-RevId: 897981232

Author: Google-ML-Automation

src/maxtext/configs/base.yml

@@ -253,7 +253,7 @@ use_2d_fsdp_sharding: False
 # deepseek moe
 base_moe_mlp_dim: 7168 # intermediate dimension at MoE layer. For a fully MoE model, base_mlp_dim must be equal to base_moe_mlp_dim.
 first_num_dense_layers: 0 # number of initial dense layers in the model
-shared_experts: 1
+shared_experts: 0
 routed_scaling_factor: 1.0 # scaling factor for routing scores
 routed_score_func: "" # scoring function for routing
 routed_bias: False # a flag if a learnable bias is added for routing

src/maxtext/configs/types.py

@@ -752,7 +752,7 @@ class DeepSeekMoE(BaseModel):
 
   base_moe_mlp_dim: int = Field(7168, description="Intermediate dimension at MoE layer (DeepSeek style).")
   first_num_dense_layers: NonNegativeInt = Field(0, description="Number of initial dense layers in the model.")
-  shared_experts: PositiveInt = Field(1, description="Number of shared experts.")
+  shared_experts: NonNegativeInt = Field(0, description="Number of shared experts.")
   routed_scaling_factor: float = Field(1.0, description="Scaling factor for routing scores.")
   routed_score_func: str = Field("", description="Scoring function for routing (e.g., 'softmax', 'sigmoid').")
   routed_bias: bool = Field(False, description="Whether to add a bias term for routing.")

src/maxtext/utils/maxtext_utils.py

@@ -217,19 +217,18 @@ def calculate_gemma2_tflops_training_per_device(config, total_ffn_flops, qkv_flo
   Calculate training TFLOP for Gemma2 as in Gemma2 we combine [local_attention, global_attention] into one decoder
   layer and we use sliding window attention in local_attention
   """
-  noncausal_attention_flops = (
-      # global attention
-      4 * config.per_device_batch_size * config.max_target_length**2 * config.num_query_heads * config.head_dim
-      +
-      # local attention
+  window = min(config.sliding_window_size, config.max_target_length)
+  global_causal_flops = (
+      2 * config.per_device_batch_size * config.max_target_length**2 * config.num_query_heads * config.head_dim
+  )
+  local_causal_flops = (
       4
       * config.per_device_batch_size
-      * config.max_target_length
-      * min(config.sliding_window_size, config.max_target_length)
+      * (config.max_target_length * window - 0.5 * window**2)
       * config.num_query_heads
       * config.head_dim
   )
-  causal_attention_flops = noncausal_attention_flops / 2
+  causal_attention_flops = global_causal_flops + local_causal_flops
   attention_tflops = causal_attention_flops * config.num_decoder_layers * 3 / 10**12
 
   # multiply num_decoder_layers by 2 because we combine [local_attention, global_attention] into one decoder layer
@@ -241,7 +240,7 @@ def calculate_gemma2_tflops_training_per_device(config, total_ffn_flops, qkv_flo
 
 
 def calculate_mixed_attention_model_tflops_training_per_device(
-    config, total_ffn_flops, qkv_flops, projection_flops, embedding_flops, attention_pattern_length
+    config, total_ffn_flops_all_layers, qkv_flops, projection_flops, embedding_flops, attention_pattern_length
 ):
   """
   Calculate training TFLOPs for models with a mixed attention pattern of local
@@ -252,34 +251,125 @@ def calculate_mixed_attention_model_tflops_training_per_device(
   num_global_layers = num_layers // attention_pattern_length
   num_local_layers = num_layers - num_global_layers
 
-  # FLOPs for a single global attention layer (full attention)
-  # Formula: 4 * batch_size * seq_len^2 * num_heads * head_dim
-  global_attention_flops_per_layer = (
-      4 * config.per_device_batch_size * config.max_target_length**2 * config.num_query_heads * config.head_dim
+  # Global causal attention uses a multiplier of 2 (instead of 4 for non-causal)
+  # since we only compute the lower triangular half of the attention matrix.
+  global_causal_flops_per_layer = (
+      2 * config.per_device_batch_size * config.max_target_length**2 * config.num_query_heads * config.head_dim
   )
 
-  # FLOPs for a single local attention layer (sliding window)
-  # Formula: 4 * batch_size * seq_len * window_size * num_heads * head_dim
-  local_attention_flops_per_layer = (
+  # Local sliding window attention directly computes the exact causal interactions
+  # via the formula `(T * W - 0.5 * W^2)`. Therefore, we use the base multiplier of 4.
+  window = min(config.sliding_window_size, config.max_target_length)
+  local_causal_flops_per_layer = (
       4
       * config.per_device_batch_size
-      * config.max_target_length
-      * min(config.sliding_window_size, config.max_target_length)
+      * (config.max_target_length * window - 0.5 * window**2)
       * config.num_query_heads
       * config.head_dim
   )
 
-  # Total attention FLOPs = (num_global_layers * FLOPs_per_global) + (num_local_layers * FLOPs_per_local)
-  noncausal_attention_flops = (
-      num_global_layers * global_attention_flops_per_layer + num_local_layers * local_attention_flops_per_layer
+  causal_attention_flops = (
+      num_global_layers * global_causal_flops_per_layer + num_local_layers * local_causal_flops_per_layer
+  )
+
+  # Convert to TFLOPs and multiply by 3 for fwd/bwd pass
+  attention_tflops = causal_attention_flops * 3 / 10**12
+
+  total_learnable_flops = total_ffn_flops_all_layers
+
+  total_learnable_flops += (qkv_flops + projection_flops) * num_layers + embedding_flops
+
+  learnable_weight_tflops = total_learnable_flops * 3 / 10**12
+
+  return attention_tflops, learnable_weight_tflops
+
+
+def calculate_gemma4_tflops_training_per_device(
+    config, total_ffn_flops_all_layers, embedding_flops, attention_pattern_length
+):
+  """
+  Calculate training TFLOPs for Gemma 4.
+  Gemma 4 has specific quirks:
+  - Different QKV projection sizes for local vs. global layers.
+  - Global-only KV sharing and varying global head dimensions.
+  """
+  num_layers = config.num_decoder_layers
+
+  num_global_layers = num_layers // attention_pattern_length
+  num_local_layers = num_layers - num_global_layers
+
+  kv_multiplier = 1 if config.share_kv_projections else 2
+  global_head_dim = config.global_head_dim or config.head_dim
+  global_num_kv_heads = config.global_num_kv_heads or config.num_kv_heads
+
+  # Global causal attention uses a multiplier of 2 (instead of 4 for non-causal)
+  # since we only compute the lower triangular half of the attention matrix.
+  global_causal_flops_per_layer = (
+      2 * config.per_device_batch_size * config.max_target_length**2 * config.num_query_heads * global_head_dim
+  )
+
+  # Local sliding window attention directly computes the exact causal interactions
+  # via the formula `(T * W - 0.5 * W^2)`. Therefore, we use the base multiplier of 4.
+  window = min(config.sliding_window_size, config.max_target_length)
+  local_causal_flops_per_layer = (
+      4
+      * config.per_device_batch_size
+      * (config.max_target_length * window - 0.5 * window**2)
+      * config.num_query_heads
+      * config.head_dim
+  )
+
+  causal_attention_flops = (
+      num_global_layers * global_causal_flops_per_layer + num_local_layers * local_causal_flops_per_layer
   )
-  causal_attention_flops = noncausal_attention_flops / 2
 
   # Convert to TFLOPs and multiply by 3 for fwd/bwd pass
   attention_tflops = causal_attention_flops * 3 / 10**12
 
-  # Learnable weights (FFN, QKV, Projections) are present in every layer.
-  learnable_weight_tflops = ((total_ffn_flops + qkv_flops + projection_flops) * num_layers + embedding_flops) * 3 / 10**12
+  global_qkv_flops_per_layer = (
+      2
+      * config.per_device_batch_size
+      * config.max_target_length
+      * config.emb_dim
+      * (config.num_query_heads + kv_multiplier * global_num_kv_heads)
+      * global_head_dim
+  )
+  global_projection_flops_per_layer = (
+      2
+      * config.per_device_batch_size
+      * config.max_target_length
+      * config.emb_dim
+      * config.num_query_heads
+      * global_head_dim
+  )
+
+  # Local layers never share KV projections (kv_multiplier is always 2).
+  local_qkv_flops_per_layer = (
+      2
+      * config.per_device_batch_size
+      * config.max_target_length
+      * config.emb_dim
+      * (config.num_query_heads + 2 * config.num_kv_heads)
+      * config.head_dim
+  )
+  local_projection_flops_per_layer = (
+      2
+      * config.per_device_batch_size
+      * config.max_target_length
+      * config.emb_dim
+      * config.num_query_heads
+      * config.head_dim
+  )
+
+  total_learnable_flops = total_ffn_flops_all_layers
+
+  total_learnable_flops += (
+      (local_qkv_flops_per_layer + local_projection_flops_per_layer) * num_local_layers
+      + (global_qkv_flops_per_layer + global_projection_flops_per_layer) * num_global_layers
+      + embedding_flops
+  )
+
+  learnable_weight_tflops = total_learnable_flops * 3 / 10**12
 
   return attention_tflops, learnable_weight_tflops
 
@@ -496,11 +586,19 @@ def get_dense_moe_layers(config):
   elif config.decoder_block == DecoderBlockType.LLAMA4:
     num_moe_layers = config.num_decoder_layers // config.interleave_moe_layer_step
     num_dense_layers = config.num_decoder_layers - num_moe_layers
+    return num_dense_layers, num_moe_layers
   elif config.decoder_block == DecoderBlockType.QWEN3_NEXT:
+    return 0, config.num_decoder_layers
+  elif config.decoder_block == DecoderBlockType.DEFAULT:
+    raise ValueError("Unsupported decoder block for dense/MoE layer calculation")
+
+  num_experts = getattr(config, "num_experts", 0)
+  if num_experts > 1:
     num_moe_layers = config.num_decoder_layers
     num_dense_layers = 0
   else:
-    raise ValueError("Currently we only support DeepSeek, Llama4, and Qwen3-Next calculation.")
+    num_moe_layers = 0
+    num_dense_layers = config.num_decoder_layers
 
   return num_dense_layers, num_moe_layers
 
@@ -601,6 +699,7 @@ def calculate_gemma3_vision_layers_tflops_per_device(config):
     learnable_weight_flops += 2 * vision_embedder_flops  # only projector is learnable, add fwd+optimizer
   else:
     learnable_weight_flops *= 3  # multiply by 3 for fwd + bwd + optimizer
+    total_attn_flops *= 3  # multiply by 3 for fwd + bwd pass
 
   # Convert to TFLOPs
   learnable_weight_tflops = learnable_weight_flops / 1e12
@@ -663,6 +762,7 @@ def calculate_llama4_vision_layers_tflops_per_device(config):
     learnable_weight_flops += 2 * projector_flops  # only projector is learnable, add fwd+optimizer
   else:
     learnable_weight_flops *= 3  # multiply by 3 for fwd + bwd + optimizer
+    total_attn_flops *= 3  # multiply by 3 for fwd + bwd pass
 
   # Convert to TFLOPs
   learnable_weight_tflops = learnable_weight_flops / 1e12
@@ -726,28 +826,40 @@ def calculate_vision_encoder_tflops(config):
 def calculate_tflops_training_per_device(config, log=True):
   """Calculate training TFLOP"""
   # MLP flops
+  is_ffn_flops_already_total = False
   if config.num_experts > 1:
     # calculation based on dropless implementation
-    if config.decoder_block in (DecoderBlockType.DEEPSEEK, DecoderBlockType.LLAMA4, DecoderBlockType.QWEN3_NEXT):
+    if config.decoder_block in (
+        DecoderBlockType.DEEPSEEK,
+        DecoderBlockType.LLAMA4,
+        DecoderBlockType.QWEN3_NEXT,
+        DecoderBlockType.GEMMA4,
+    ):
       total_ffn_flops = calculate_routed_and_shared_ffn_tflops_per_device(config)
+      is_ffn_flops_already_total = True
     else:
       gate_flops = 2 * config.per_device_batch_size * config.max_target_length * config.emb_dim * config.num_experts
       total_ffn_flops = (
-          gate_flops + calculate_ffn_mamtul_tflops_per_device(config, config.mlp_dim) * config.num_experts_per_tok
+          gate_flops + calculate_ffn_mamtul_tflops_per_device(config, config.moe_mlp_dim) * config.num_experts_per_tok
       )
   else:
     total_ffn_flops = calculate_ffn_mamtul_tflops_per_device(config, config.mlp_dim)
 
+  total_ffn_flops_all_layers = (
+      total_ffn_flops if is_ffn_flops_already_total else total_ffn_flops * config.num_decoder_layers
+  )
+
   # Attention flops
   if config.attention_type == "mla":
     qkv_flops, causal_attention_flops, projection_flops = calculate_mla_tflops_per_device(config)
   else:
+    kv_multiplier = 1 if config.share_kv_projections else 2
     qkv_flops = (
         2
         * config.per_device_batch_size
         * config.max_target_length
         * config.emb_dim
-        * (config.num_query_heads + 2 * config.num_kv_heads)
+        * (config.num_query_heads + kv_multiplier * config.num_kv_heads)
         * config.head_dim
     )
     noncausal_attention_flops = (
@@ -768,7 +880,8 @@ def calculate_tflops_training_per_device(config, log=True):
     # NVIDIA/NeMo (2025 April): https://github.com/NVIDIA/NeMo/blob/ba4d6d116463de512ff0cfc14641aa6cf4577a42/nemo/utils/flops_formulas.py#L259-L272
     causal_attention_flops = noncausal_attention_flops / 2
 
-  # Embedding flops
+  # Embedding flops (counts only the unembedding projection; the embedding lookup is a gather operation
+  # that performs no dense math, matching standard MFU hardware calculations)
   embedding_flops = 2 * config.per_device_batch_size * config.max_target_length * config.emb_dim * config.vocab_size
 
   # Combine flops with number of decoder layers
@@ -778,26 +891,30 @@ def calculate_tflops_training_per_device(config, log=True):
     )
   elif config.decoder_block == DecoderBlockType.GEMMA3:
     attention_tflops, learnable_weight_tflops = calculate_mixed_attention_model_tflops_training_per_device(
-        config, total_ffn_flops, qkv_flops, projection_flops, embedding_flops, attention_pattern_length=6
+        config, total_ffn_flops_all_layers, qkv_flops, projection_flops, embedding_flops, attention_pattern_length=6
     )
   elif config.decoder_block == DecoderBlockType.GPT_OSS:
     attention_tflops, learnable_weight_tflops = calculate_mixed_attention_model_tflops_training_per_device(
-        config, total_ffn_flops, qkv_flops, projection_flops, embedding_flops, attention_pattern_length=2
+        config, total_ffn_flops_all_layers, qkv_flops, projection_flops, embedding_flops, attention_pattern_length=2
     )
   elif config.decoder_block == DecoderBlockType.LLAMA4:
     # Use the new helper to calculate attention TFLOPs correctly.
     attention_tflops = calculate_llama4_attention_tflops(config)
     # The learnable weight calculation remains the same as it correctly handles Llama4's MoE structure.
     learnable_weight_tflops = (
-        (total_ffn_flops + (qkv_flops + projection_flops) * config.num_decoder_layers + embedding_flops) * 3 / 10**12
+        (total_ffn_flops_all_layers + (qkv_flops + projection_flops) * config.num_decoder_layers + embedding_flops)
+        * 3
+        / 10**12
     )
   elif config.decoder_block == DecoderBlockType.GEMMA4:
-    attention_tflops, learnable_weight_tflops = calculate_mixed_attention_model_tflops_training_per_device(
-        config, total_ffn_flops, qkv_flops, projection_flops, embedding_flops, attention_pattern_length=6
+    attention_tflops, learnable_weight_tflops = calculate_gemma4_tflops_training_per_device(
+        config, total_ffn_flops_all_layers, embedding_flops, attention_pattern_length=6
     )
   elif config.decoder_block == DecoderBlockType.DEEPSEEK:
     learnable_weight_tflops = (
-        (total_ffn_flops + (qkv_flops + projection_flops) * config.num_decoder_layers + embedding_flops) * 3 / 10**12
+        (total_ffn_flops_all_layers + (qkv_flops + projection_flops) * config.num_decoder_layers + embedding_flops)
+        * 3
+        / 10**12
     )
     attention_tflops = causal_attention_flops * config.num_decoder_layers * 3 / 10**12
   elif config.decoder_block == DecoderBlockType.QWEN3_NEXT:
@@ -808,7 +925,7 @@ def calculate_tflops_training_per_device(config, log=True):
 
     # Weights TFLOPs:
     total_weights = (
-        total_ffn_flops
+        total_ffn_flops_all_layers
         + embedding_flops
         + (qkv_flops + projection_flops) * num_full_attn_layers
         + gdn_weight_flops_per_layer * num_linear_attn_layers
@@ -821,7 +938,9 @@ def calculate_tflops_training_per_device(config, log=True):
   else:
     # multiply by 3 for both feed forward and back propagation flops
     learnable_weight_tflops = (
-        ((total_ffn_flops + qkv_flops + projection_flops) * config.num_decoder_layers + embedding_flops) * 3 / 10**12
+        (total_ffn_flops_all_layers + (qkv_flops + projection_flops) * config.num_decoder_layers + embedding_flops)
+        * 3
+        / 10**12
     )
     attention_tflops = causal_attention_flops * config.num_decoder_layers * 3 / 10**12
 

tests/integration/smoke/train_using_ragged_dot_smoke_test.py

@@ -55,6 +55,7 @@ def test_tiny_config(self, quantization: str):
             "decoder_block=deepseek",
             "attention_type=mla",
             "num_experts=2",
+            "shared_experts=1",
             # Enable sparse_matmul.
             "sparse_matmul=True",
             # Enable ragged_dot.