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

make linter happy

Browse source
This commit is contained in:
slaren 2024-04-02 18:21:45 +02:00
parent d08a1f4860
commit 9530398013
1 changed files with 9 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

18
convert.py
View file

@ -1257,19 +1257,19 @@ def convert_model_names(model: LazyModel, params: Params, skip_unknown: bool) ->
# merge experts into one tensor # merge experts into one tensor
if params.n_experts > 0: if params.n_experts > 0:
for l in range(params.n_layer): for i_l in range(params.n_layer):
for w in range(1, 4): for w in range(1, 4):
experts = [] experts = []
for e in range(params.n_experts): for e in range(params.n_experts):
if f"layers.{l}.feed_forward.experts.{e}.w{w}.weight" in model: if f"layers.{i_l}.feed_forward.experts.{e}.w{w}.weight" in model:
experts.append(model[f"layers.{l}.feed_forward.experts.{e}.w{w}.weight"]) experts.append(model[f"layers.{i_l}.feed_forward.experts.{e}.w{w}.weight"])
del tmp[f"layers.{l}.feed_forward.experts.{e}.w{w}.weight"] del tmp[f"layers.{i_l}.feed_forward.experts.{e}.w{w}.weight"]
elif f"model.layers.{l}.block_sparse_moe.experts.{e}.w{w}.weight" in model: elif f"model.layers.{i_l}.block_sparse_moe.experts.{e}.w{w}.weight" in model:
experts.append(model[f"model.layers.{l}.block_sparse_moe.experts.{e}.w{w}.weight"]) experts.append(model[f"model.layers.{i_l}.block_sparse_moe.experts.{e}.w{w}.weight"])
del tmp[f"model.layers.{l}.block_sparse_moe.experts.{e}.w{w}.weight"] del tmp[f"model.layers.{i_l}.block_sparse_moe.experts.{e}.w{w}.weight"]
else: else:
raise ValueError(f"Expert tensor not found: layers.{l}.feed_forward.experts.{e}.w{w}.weight") raise ValueError(f"Expert tensor not found: layers.{i_l}.feed_forward.experts.{e}.w{w}.weight")
tmp[f"layers.{l}.feed_forward.experts.w{w}.weight"] = pack_experts_lazy(experts) tmp[f"layers.{i_l}.feed_forward.experts.w{w}.weight"] = pack_experts_lazy(experts)
# HF models permut or pack some of the tensors, so we need to undo that # HF models permut or pack some of the tensors, so we need to undo that
for i in itertools.count(): for i in itertools.count():