rwkv7: do not quantize small yet 2D lora weights
Signed-off-by: Molly Sophia <mollysophia379@gmail.com>
This commit is contained in:
parent
41a80dfb03
commit
1a9c2635b3
1 changed files with 10 additions and 1 deletions
|
@ -756,10 +756,19 @@ static void llama_model_quantize_impl(const std::string & fname_inp, const std::
|
|||
// NOTE: can't use LLM_TN here because the layer number is not known
|
||||
quantize &= name.find("ssm_conv1d.weight") == std::string::npos;
|
||||
|
||||
// do not quantize RWKV's time_mix_first tensors
|
||||
// do not quantize RWKV's small yet 2D weights
|
||||
quantize &= name.find("time_mix_first.weight") == std::string::npos;
|
||||
quantize &= name.find("time_mix_w0.weight") == std::string::npos;
|
||||
quantize &= name.find("time_mix_w1.weight") == std::string::npos;
|
||||
quantize &= name.find("time_mix_w2.weight") == std::string::npos;
|
||||
quantize &= name.find("time_mix_v0.weight") == std::string::npos;
|
||||
quantize &= name.find("time_mix_v1.weight") == std::string::npos;
|
||||
quantize &= name.find("time_mix_v2.weight") == std::string::npos;
|
||||
quantize &= name.find("time_mix_a0.weight") == std::string::npos;
|
||||
quantize &= name.find("time_mix_a1.weight") == std::string::npos;
|
||||
quantize &= name.find("time_mix_a2.weight") == std::string::npos;
|
||||
quantize &= name.find("time_mix_g1.weight") == std::string::npos;
|
||||
quantize &= name.find("time_mix_g2.weight") == std::string::npos;
|
||||
quantize &= name.find("time_mix_decay_w1.weight") == std::string::npos;
|
||||
quantize &= name.find("time_mix_decay_w2.weight") == std::string::npos;
|
||||
quantize &= name.find("time_mix_lerp_fused.weight") == std::string::npos;
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue