happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

f16 gguf cannot run without context length

This commit is contained in:
Yee Man Chan 2026-02-05 07:44:10 +08:00
parent 82215a053b
commit a82103eb48
1 changed files with 9 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
convert_hf_to_gguf.py
View File

@ -5086,6 +5086,15 @@ class KimiLinearModel(TextModel):
super().set_gguf_parameters() super().set_gguf_parameters()
self.gguf_writer.add_vocab_size(self.hparams["vocab_size"]) self.gguf_writer.add_vocab_size(self.hparams["vocab_size"])
# Use find_hparam for context length
# Kimi uses model_max_length
if (n_ctx := self.find_hparam(["max_position_embeddings", "model_max_length", "n_ctx", "n_positions"], optional=True)) is not None:
self.gguf_writer.add_context_length(n_ctx)
else:
# Default to 4096 if not found
logger.warning("No context length found in config, defaulting to 4096")
self.gguf_writer.add_context_length(4096)
# KDA & MLA params # KDA & MLA params
# Get ssm_d_conv from linear_attn_config.short_conv_kernel_size or ssm_d_conv # Get ssm_d_conv from linear_attn_config.short_conv_kernel_size or ssm_d_conv
linear_attn_config = self.hparams["linear_attn_config"] linear_attn_config = self.hparams["linear_attn_config"]