vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Use Q8_0 quantization from gguf module.

Browse source 
This makes tensors exactly as in https://huggingface.co/Arki05/Grok-1-GGUF/tree/main/Q8_0
This commit is contained in:
Heiner 2024-05-23 11:10:59 +02:00
parent f177b6596c
commit e2f13a3346
1 changed files with 6 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

11
convert_grok.py
View file

@ -124,7 +124,7 @@ def get_weights(fn):
def quantize_q8_0(tensor: torch.Tensor) -> torch.CharTensor: def quantize_q8_0(tensor: torch.Tensor) -> torch.CharTensor:
# equivalent to ggml_quantize_q8_0 in ggml.c # equivalent to ggml_quantize_q8_0 in ggml.c (modulo rounding away from zero)
assert tensor.shape[1] % GGML_QK8_0 == 0 assert tensor.shape[1] % GGML_QK8_0 == 0
tensor = tensor.reshape(-1, GGML_QK8_0) tensor = tensor.reshape(-1, GGML_QK8_0)
scale = tensor.abs().max(dim=-1, keepdim=True).values / ((1 << 7) - 1) scale = tensor.abs().max(dim=-1, keepdim=True).values / ((1 << 7) - 1)
@ -135,7 +135,7 @@ def quantize_q8_0(tensor: torch.Tensor) -> torch.CharTensor:
def quantize_q4_0(tensor: torch.Tensor) -> torch.CharTensor: def quantize_q4_0(tensor: torch.Tensor) -> torch.CharTensor:
# equivalent to ggml_quantize_q4_0 in ggml.c # equivalent to ggml_quantize_q4_0 in ggml.c (modulo rounding away from zero)
assert tensor.shape[1] % GGML_QK4_0 == 0 assert tensor.shape[1] % GGML_QK4_0 == 0
tensor = tensor.reshape(-1, GGML_QK4_0) tensor = tensor.reshape(-1, GGML_QK4_0)
abs_max_indices = tensor.abs().max(dim=-1, keepdim=True).indices abs_max_indices = tensor.abs().max(dim=-1, keepdim=True).indices
@ -150,7 +150,7 @@ def quantize_q4_0(tensor: torch.Tensor) -> torch.CharTensor:
def quantize_q4_1(tensor: torch.Tensor) -> torch.CharTensor: def quantize_q4_1(tensor: torch.Tensor) -> torch.CharTensor:
# equivalent to ggml_quantize_q4_1 in ggml.c # equivalent to ggml_quantize_q4_1 in ggml.c (modulo rounding away from zero)
assert tensor.shape[1] % GGML_QK4_1 == 0 assert tensor.shape[1] % GGML_QK4_1 == 0
tensor = tensor.reshape(-1, GGML_QK4_1) tensor = tensor.reshape(-1, GGML_QK4_1)
abs_max_indices = tensor.max(dim=-1, keepdim=True).indices abs_max_indices = tensor.max(dim=-1, keepdim=True).indices
@ -170,13 +170,14 @@ def quantize_q4_1(tensor: torch.Tensor) -> torch.CharTensor:
def maybe_quantize_tensor(tensor, ggml_type): def maybe_quantize_tensor(tensor, ggml_type):
assert tensor.dtype == torch.float32 assert tensor.dtype == torch.float32
if ggml_type == gguf.GGMLQuantizationType.F32: if ggml_type == gguf.GGMLQuantizationType.F32:
return tensor.float() return tensor.float()
elif ggml_type == gguf.GGMLQuantizationType.F16: elif ggml_type == gguf.GGMLQuantizationType.F16:
return tensor.half() return tensor.half()
elif ggml_type == gguf.GGMLQuantizationType.Q8_0: elif ggml_type == gguf.GGMLQuantizationType.Q8_0:
return quantize_q8_0(tensor) if tensor.device.type == "meta":
return quantize_q8_0(tensor) # Cannot convert into numpy array.
return torch.from_numpy(gguf.quantize_q8_0(tensor.numpy()))
elif ggml_type == gguf.GGMLQuantizationType.Q4_0: elif ggml_type == gguf.GGMLQuantizationType.Q4_0:
return quantize_q4_0(tensor) return quantize_q4_0(tensor)
elif ggml_type == gguf.GGMLQuantizationType.Q4_1: elif ggml_type == gguf.GGMLQuantizationType.Q4_1: