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

*.py: Convert logger error and sys.exit() into a raise exception (for atypical error)

Browse source
This commit is contained in:
brian khuu 2024-04-18 11:13:34 +10:00
parent cf38b4b831
commit dc798d23d7
2 changed files with 30 additions and 57 deletions
Download patch file Download diff file Expand all files Collapse all files

83
convert-hf-to-gguf.py
View file

@ -154,8 +154,7 @@ class Model(ABC):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -572,8 +571,7 @@ class BloomModel(Model):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -656,8 +654,7 @@ class MPTModel(Model):
else: else:
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -700,8 +697,7 @@ class OrionModel(Model):
elif "model_max_length" in self.hparams: elif "model_max_length" in self.hparams:
ctx_length = self.hparams["model_max_length"] ctx_length = self.hparams["model_max_length"]
else: else:
logger.error("gguf: can not find ctx length parameter.") raise ValueError("gguf: can not find ctx length parameter.")
sys.exit()
self.gguf_writer.add_file_type(self.ftype) self.gguf_writer.add_file_type(self.ftype)
self.gguf_writer.add_name(self.dir_model.name) self.gguf_writer.add_name(self.dir_model.name)
@ -739,8 +735,7 @@ class OrionModel(Model):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -782,8 +777,7 @@ class BaichuanModel(Model):
elif "model_max_length" in self.hparams: elif "model_max_length" in self.hparams:
ctx_length = self.hparams["model_max_length"] ctx_length = self.hparams["model_max_length"]
else: else:
logger.error("gguf: can not find ctx length parameter.") raise ValueError("gguf: can not find ctx length parameter.")
sys.exit()
self.gguf_writer.add_name(self.dir_model.name) self.gguf_writer.add_name(self.dir_model.name)
self.gguf_writer.add_source_hf_repo(hf_repo) self.gguf_writer.add_source_hf_repo(hf_repo)
@ -837,8 +831,7 @@ class BaichuanModel(Model):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -940,8 +933,7 @@ class XverseModel(Model):
elif "model_max_length" in self.hparams: elif "model_max_length" in self.hparams:
ctx_length = self.hparams["model_max_length"] ctx_length = self.hparams["model_max_length"]
else: else:
logger.error("gguf: can not find ctx length parameter.") raise ValueError("gguf: can not find ctx length parameter.")
sys.exit()
self.gguf_writer.add_name(self.dir_model.name) self.gguf_writer.add_name(self.dir_model.name)
self.gguf_writer.add_source_hf_repo(hf_repo) self.gguf_writer.add_source_hf_repo(hf_repo)
@ -990,8 +982,7 @@ class XverseModel(Model):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -1095,8 +1086,7 @@ class FalconModel(Model):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -1198,10 +1188,9 @@ class RefactModel(Model):
data = data_torch.squeeze().numpy() data = data_torch.squeeze().numpy()
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight",)) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -1267,8 +1256,7 @@ class PersimmonModel(Model):
data = data_torch.to(torch.float32).squeeze().numpy() data = data_torch.to(torch.float32).squeeze().numpy()
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
logger.info(f"{new_name}, n_dims = {n_dims}, {old_dtype} --> {data.dtype}") logger.info(f"{new_name}, n_dims = {n_dims}, {old_dtype} --> {data.dtype}")
self.gguf_writer.add_tensor(new_name, data) self.gguf_writer.add_tensor(new_name, data)
@ -1483,8 +1471,7 @@ class LlamaModel(Model):
new_name = tensor_map.get_name(merged_name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(merged_name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
logger.info(f"{new_name}, n_dims = {len(data.shape)}, shape = {data.shape} --> {data.dtype}") logger.info(f"{new_name}, n_dims = {len(data.shape)}, shape = {data.shape} --> {data.dtype}")
@ -1494,8 +1481,7 @@ class LlamaModel(Model):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -1587,8 +1573,7 @@ class GrokModel(Model):
new_name = tensor_map.get_name(merged_name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(merged_name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
logger.info(f"{new_name}, n_dims = {len(data.shape)}, shape = {data.shape} --> {data.dtype}") logger.info(f"{new_name}, n_dims = {len(data.shape)}, shape = {data.shape} --> {data.dtype}")
@ -1598,8 +1583,7 @@ class GrokModel(Model):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -1692,8 +1676,7 @@ class DbrxModel(Model):
# https://huggingface.co/databricks/dbrx-instruct/blob/main/model.safetensors.index.json#L15 # https://huggingface.co/databricks/dbrx-instruct/blob/main/model.safetensors.index.json#L15
new_name = tensor_map.get_name(name if not experts else name + ".weight", try_suffixes=(".weight",)) new_name = tensor_map.get_name(name if not experts else name + ".weight", try_suffixes=(".weight",))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -1701,8 +1684,7 @@ class DbrxModel(Model):
# Most of the codebase that takes in 1D tensors only handles F32 tensors # Most of the codebase that takes in 1D tensors only handles F32 tensors
# and most of the outputs tensors are F32. # and most of the outputs tensors are F32.
if data_dtype != np.float32 and n_dims == 1: if data_dtype != np.float32 and n_dims == 1:
logger.error(f"Can not map tensor {name!r}: all 1D tensors must be F32") raise ValueError(f"Can not map tensor {name!r}: all 1D tensors must be F32")
sys.exit()
# if f32 desired, convert any float16 to float32 # if f32 desired, convert any float16 to float32
if self.ftype == 0 and data_dtype == np.float16: if self.ftype == 0 and data_dtype == np.float16:
@ -1774,8 +1756,7 @@ class MiniCPMModel(Model):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -1858,8 +1839,7 @@ class QwenModel(Model):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -2027,8 +2007,7 @@ class GPT2Model(Model):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -2211,8 +2190,7 @@ class PlamoModel(Model):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
# shuffle for broadcasting of gqa in ggml_mul_mat # shuffle for broadcasting of gqa in ggml_mul_mat
if new_name.endswith("attn_q.weight"): if new_name.endswith("attn_q.weight"):
@ -2289,8 +2267,7 @@ class CodeShellModel(Model):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -2438,8 +2415,7 @@ in chat mode so that the conversation can end normally.")
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -2567,8 +2543,7 @@ class BertModel(Model):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
# convert any unsupported data types to float32 # convert any unsupported data types to float32
if data_torch.dtype not in (torch.float16, torch.float32): if data_torch.dtype not in (torch.float16, torch.float32):
@ -2684,8 +2659,7 @@ class GemmaModel(Model):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
data_dtype = data.dtype data_dtype = data.dtype
@ -2796,8 +2770,7 @@ class MambaModel(Model):
# map tensor names # map tensor names
new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes=(".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor {name!r}") raise ValueError(f"Can not map tensor {name!r}")
sys.exit()
if name.endswith(".A_log"): if name.endswith(".A_log"):
logger.debug("A_log --> A ==> " + new_name) logger.debug("A_log --> A ==> " + new_name)

4
convert-persimmon-to-gguf.py
View file

@ -122,8 +122,8 @@ def main():
data = data_torch.to(torch.float32).squeeze().numpy() data = data_torch.to(torch.float32).squeeze().numpy()
new_name = tensor_map.get_name(name, try_suffixes = (".weight", ".bias")) new_name = tensor_map.get_name(name, try_suffixes = (".weight", ".bias"))
if new_name is None: if new_name is None:
logger.error(f"Can not map tensor '{name}'") raise ValueError(f"Can not map tensor '{name}'")
sys.exit()
n_dims = len(data.shape) n_dims = len(data.shape)
logger.debug(f"{new_name}, n_dims = {str(n_dims)}, {str(old_dtype)} --> {str(data.dtype)}") logger.debug(f"{new_name}, n_dims = {str(n_dims)}, {str(old_dtype)} --> {str(data.dtype)}")
gguf_writer.add_tensor(new_name, data) gguf_writer.add_tensor(new_name, data)