rebase and signing

Signed-off-by: ryan-mangeno <[email protected]>

Author: ryan-mangeno

convert_hf_to_gguf_update.py

@@ -130,7 +130,6 @@ class TOKENIZER_TYPE(IntEnum):
     {"name": "seed-coder",       "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/ByteDance-Seed/Seed-Coder-8B-Base", },
     {"name": "smoldocling",      "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/ds4sd/SmolDocling-256M-preview", },
     {"name": "a.x-4.0",          "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/skt/A.X-4.0", },
-    {"name": "smoldocling",      "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/ds4sd/SmolDocling-256M-preview", },
 ]
 
 # some models are known to be broken upstream, so we will skip them as exceptions

gguf-py/gguf/tensor_mapping.py

@@ -359,6 +359,7 @@ class TensorNameMap:
             "transformer.h.{bid}.mlp.c_fc_1",                         # exaone
             "model.layers.{bid}.feed_forward.up_proj",                # llama4
             "transformer_encoder.{bid}.ffn.w12",                      # neobert
+            "blk.{bid}.ffn_up",                                       # smoldocling
             "blk.{bid}.ffn_up",                                       # smoldocling                                     
         ),
 

src/llama-vocab.cpp

@@ -1669,6 +1669,9 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
             } else if (
                 tokenizer_pre == "hunyuan") {
                 pre_type = LLAMA_VOCAB_PRE_TYPE_HUNYUAN;
+            } else if (
+                tokenizer_pre == "smoldocling") {
+                pre_type = LLAMA_VOCAB_PRE_TYPE_SMOLDOCLING;
                 clean_spaces = false;
             } else if (
                 tokenizer_pre == "smoldocling") {