model : support for LlamaBidirectionalModel architecture (#18220)

* model: llama-embed-nemotron

* minor: python lint

* changed arch-name

* templated llm_build_llama to be used for both llama and llama-embed arch

gguf-py/gguf/constants.py

@@ -449,6 +449,7 @@ class MODEL_ARCH(IntEnum):
     RND1             = auto()
     PANGU_EMBED      = auto()
     MISTRAL3         = auto()
+    LLAMA_EMBED      = auto()
 
 
 class VISION_PROJECTOR_TYPE(IntEnum):
@@ -844,6 +845,7 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
     MODEL_ARCH.RND1:             "rnd1",
     MODEL_ARCH.PANGU_EMBED:      "pangu-embedded",
     MODEL_ARCH.MISTRAL3:         "mistral3",
+    MODEL_ARCH.LLAMA_EMBED:      "llama-embed",
 }
 
 VISION_PROJECTOR_TYPE_NAMES: dict[VISION_PROJECTOR_TYPE, str] = {
@@ -3196,6 +3198,26 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
         MODEL_TENSOR.FFN_DOWN_EXP,
         MODEL_TENSOR.FFN_UP_EXP,
     ],
+    MODEL_ARCH.LLAMA_EMBED: [
+        MODEL_TENSOR.TOKEN_EMBD,
+        MODEL_TENSOR.OUTPUT_NORM,
+        MODEL_TENSOR.OUTPUT,
+        MODEL_TENSOR.ROPE_FREQS,
+        MODEL_TENSOR.ATTN_NORM,
+        MODEL_TENSOR.ATTN_Q,
+        MODEL_TENSOR.ATTN_K,
+        MODEL_TENSOR.ATTN_V,
+        MODEL_TENSOR.ATTN_OUT,
+        MODEL_TENSOR.ATTN_ROT_EMBD,
+        MODEL_TENSOR.FFN_GATE_INP,
+        MODEL_TENSOR.FFN_NORM,
+        MODEL_TENSOR.FFN_GATE,
+        MODEL_TENSOR.FFN_DOWN,
+        MODEL_TENSOR.FFN_UP,
+        MODEL_TENSOR.FFN_GATE_EXP,
+        MODEL_TENSOR.FFN_DOWN_EXP,
+        MODEL_TENSOR.FFN_UP_EXP,
+    ]
     # TODO
 }