model : support MTP in GLM-4.7-Flash (#24868)

This commit is contained in:
jacekpoplawski
2026-08-03 20:27:52 +02:00
committed by GitHub
parent ee0445c99c
commit 57c092139a
4 changed files with 379 additions and 25 deletions
+7
View File
@@ -3221,6 +3221,13 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
MODEL_TENSOR.FFN_DOWN_SHEXP,
MODEL_TENSOR.FFN_UP_SHEXP,
MODEL_TENSOR.FFN_EXP_PROBS_B,
# NextN/MTP tensors
MODEL_TENSOR.NEXTN_EH_PROJ,
MODEL_TENSOR.NEXTN_EMBED_TOKENS,
MODEL_TENSOR.NEXTN_ENORM,
MODEL_TENSOR.NEXTN_HNORM,
MODEL_TENSOR.NEXTN_SHARED_HEAD_HEAD,
MODEL_TENSOR.NEXTN_SHARED_HEAD_NORM,
],
MODEL_ARCH.DEEPSEEK2OCR: [
MODEL_TENSOR.TOKEN_EMBD,