From caf309157ba62c6feb61dd587fad046aabeae50b Mon Sep 17 00:00:00 2001 From: Francis Couture-Harpin Date: Mon, 16 Sep 2024 12:05:29 -0400 Subject: [PATCH] convert : adapt MiniCPM3 to separate rope_freqs insertion MiniCPM3's tokenizer is treated as a SentencePiece tokenizer to avoid having to run its custom Python code which mixes tokenization in the same file as tool calls. gguf-py : add long and short RoPE factors to tensor mappings Empty, but the key names are used to populate the mappings. --- gguf-py/gguf/tensor_mapping.py | 3 +++ 1 file changed, 3 insertions(+) diff --git a/gguf-py/gguf/tensor_mapping.py b/gguf-py/gguf/tensor_mapping.py index 3ff70cd7..9688b02c 100644 --- a/gguf-py/gguf/tensor_mapping.py +++ b/gguf-py/gguf/tensor_mapping.py @@ -82,6 +82,9 @@ class TensorNameMap: "rope.freqs", # llama-pth "rotary_pos_emb.inv_freq", # chatglm ), + + MODEL_TENSOR.ROPE_FACTORS_LONG: (), + MODEL_TENSOR.ROPE_FACTORS_SHORT: (), } block_mappings_cfg: dict[MODEL_TENSOR, tuple[str, ...]] = {