Add support for ArcticForCausalLM (#7020)

* common : increase max number of experts to 128

* common : add tensor LLM_TENSOR_FFN_NORM_EXPS for normalization before MoE that runs in parallel to attention + ffn

* gguf-py : add architecture-specific block mappings that override selected general block mappings

* convert-hf : add model conversion support for ArcticForCausalLM

* convert-hf : use added_tokens_decoder from tokenizer_config.json to redefine tokens from SentencePiece model (only for ArcticForCausalLM)

* llama : add inference support for LLM_ARCH_ARCTIC

---------

Co-authored-by: Stanisław Szymczyk <sszymczy@gmail.com>

1 files changed, 18 insertions, 1 deletions

diff --git a/gguf-py/gguf/tensor_mapping.py b/gguf-py/gguf/tensor_mapping.py
index 8e1cac91..8b1b21d7 100644
--- a/gguf-py/gguf/tensor_mapping.py
+++ b/gguf-py/gguf/tensor_mapping.py
@@ -244,6 +244,7 @@ class TensorNameMap:
             "encoder.layers.{bid}.mlp.fc11",                          # nomic-bert
             "model.layers.{bid}.mlp.c_fc",                            # starcoder2
             "encoder.layer.{bid}.mlp.gated_layers_v",                 # jina-bert-v2
+            "model.layers.{bid}.residual_mlp.w3",                     # arctic
         ),
 
         MODEL_TENSOR.FFN_UP_EXP: (
@@ -272,6 +273,7 @@ class TensorNameMap:
             "encoder.layers.{bid}.mlp.fc12",              # nomic-bert
             "encoder.layer.{bid}.mlp.gated_layers_w",     # jina-bert-v2
             "transformer.h.{bid}.mlp.linear_1",           # refact
+            "model.layers.{bid}.residual_mlp.w1",         # arctic
         ),
 
         MODEL_TENSOR.FFN_GATE_EXP: (
@@ -306,6 +308,7 @@ class TensorNameMap:
             "encoder.layers.{bid}.mlp.fc2",                           # nomic-bert
             "model.layers.{bid}.mlp.c_proj",                          # starcoder2
             "encoder.layer.{bid}.mlp.wo",                             # jina-bert-v2
+            "model.layers.{bid}.residual_mlp.w2",                     # arctic
         ),
 
         MODEL_TENSOR.FFN_DOWN_EXP: (
@@ -382,6 +385,18 @@ class TensorNameMap:
         ),
     }
 
+    # architecture-specific block mappings
+    arch_block_mappings_cfg: dict[MODEL_ARCH, dict[MODEL_TENSOR, tuple[str, ...]]] = {
+        MODEL_ARCH.ARCTIC: {
+            MODEL_TENSOR.FFN_NORM: (
+                "model.layers.{bid}.residual_layernorm",
+            ),
+            MODEL_TENSOR.FFN_NORM_EXP: (
+                "model.layers.{bid}.post_attention_layernorm",
+            ),
+        },
+    }
+
     mapping: dict[str, tuple[MODEL_TENSOR, str]]
 
     def __init__(self, arch: MODEL_ARCH, n_blocks: int):
@@ -393,12 +408,14 @@ class TensorNameMap:
             self.mapping[tensor_name] = (tensor, tensor_name)
             for key in keys:
                 self.mapping[key] = (tensor, tensor_name)
+        if arch in self.arch_block_mappings_cfg:
+            self.block_mappings_cfg.update(self.arch_block_mappings_cfg[arch])
         for bid in range(n_blocks):
             for tensor, keys in self.block_mappings_cfg.items():
                 if tensor not in MODEL_TENSORS[arch]:
                     continue
                 # TODO: make this configurable
-                n_experts = 60
+                n_experts = 128
                 for xid in range(n_experts):
                     tensor_name = TENSOR_NAMES[tensor].format(bid = bid, xid = xid)
                     self.mapping[tensor_name] = (tensor, tensor_name)