vrr/kvcache-ai-ktransformers
2
0
Fork 0
mirror of https://github.com/kvcache-ai/ktransformers.git synced 2025-09-15 01:29:42 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

support qwen3, dont speak human language

Browse source
This commit is contained in:
djw 2025-04-28 08:44:47 +00:00
parent f3d842a0ca
commit 3f9bbf1181
30 changed files with 3696 additions and 290 deletions
Download patch file Download diff file Expand all files Collapse all files

15
ktransformers/util/custom_gguf.py
View file

@ -912,6 +912,9 @@ def translate_name_to_gguf(name):
name = name.replace(".self_attn.q_a_proj", ".attn_q_a")
name = name.replace(".self_attn.q_a_layernorm", ".attn_q_a_norm")
name = name.replace(".self_attn.q_b_proj", ".attn_q_b")
name = name.replace(".self_attn.q_norm", ".attn_q_norm")
name = name.replace(".self_attn.k_norm", ".attn_k_norm")
name = name.replace(".shared_expert.", ".shared_experts.")
name = name.replace(".shared_expert_", ".shared_experts_")
@ -922,17 +925,23 @@ def translate_name_to_gguf(name):
name = name.replace(".mlp.shared_experts.gate_proj", ".ffn_gate_shexp")
name = name.replace(".mlp.shared_experts.up_proj", ".ffn_up_shexp")
name = name.replace(".mlp.shared_experts_gate", ".ffn_gate_inp_shexp")
name = name.replace(".mlp.experts", "")
name = name.replace(".mlp.experts.ffn_down_exps", ".ffn_down_exps")
name = name.replace(".mlp.experts.ffn_gate_exps", ".ffn_gate_exps")
name = name.replace(".mlp.experts.ffn_up_exps", ".ffn_up_exps")
name = name.replace(".block_sparse_moe.gate.", ".ffn_gate_inp.")
name = name.replace(".block_sparse_moe.experts", "")
name = name.replace(".feed_forward.experts", "")
name = name.replace(".feed_forward.router", ".ffn_gate_inp")
name = name.replace(".feed_forward.shared_experts.down_proj", ".ffn_down_shexp")
name = name.replace(".feed_forward.shared_experts.gate_proj", ".ffn_gate_shexp")
name = name.replace(".feed_forward.shared_experts.up_proj", ".ffn_up_shexp")
return name
if __name__ == '__main__':
gguf_path = '/mnt/data/model/DeepSeek-Coder-V2-GGUF-WJH'
loader = GGUFLoader(gguf_path)