rixty/KoboldAI-Client
1
1
Fork 0
mirror of https://github.com/KoboldAI/KoboldAI-Client.git synced 2025-04-13 09:52:40 +02:00
Code Issues Projects Releases Wiki Activity

Nevermind, dividing by 4 is actually correct...

Browse Source
This commit is contained in:
Gnome Ann 2022-03-19 00:55:04 -04:00
parent f16b61ec77
commit 6c20d0d657
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
tpu_mtj_backend.py
View File

@ -885,7 +885,7 @@ def read_neox_checkpoint(state, path, config, checkpoint_shards=2):
original_shape = shards[0][key].shape original_shape = shards[0][key].shape
for checkpoint_shard in range(checkpoint_shards): for checkpoint_shard in range(checkpoint_shards):
if key in ("attention.dense.bias", "mlp.dense_4h_to_h.bias"): if key in ("attention.dense.bias", "mlp.dense_4h_to_h.bias"):
shards[checkpoint_shard][key] /= config["cores_per_replica"] shards[checkpoint_shard][key] /= output_shards
if key != "word_embeddings.weight" and shards[checkpoint_shard][key].ndim == 2: if key != "word_embeddings.weight" and shards[checkpoint_shard][key].ndim == 2:
shards[checkpoint_shard][key] = shards[checkpoint_shard][key].T shards[checkpoint_shard][key] = shards[checkpoint_shard][key].T
tensor = shards[checkpoint_shard][key] tensor = shards[checkpoint_shard][key]