Add convert.py.
Browse files- convert.py +26 -0
convert.py
ADDED
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from transformers import AutoModelForCausalLM
|
2 |
+
import torch
|
3 |
+
from safetensors.torch import save_file
|
4 |
+
|
5 |
+
model = AutoModelForCausalLM.from_pretrained("microsoft/Phi-3-mini-4k-instruct", trust_remote_code=True)
|
6 |
+
|
7 |
+
params = model.state_dict()
|
8 |
+
params2 = {}
|
9 |
+
|
10 |
+
for r in params.keys():
|
11 |
+
if "gate_up_proj" in r:
|
12 |
+
(gate, up) = params[r].chunk(2)
|
13 |
+
params2[r.replace("gate_up_proj", "gate_proj")] = gate
|
14 |
+
params2[r.replace("gate_up_proj", "up_proj")] = up
|
15 |
+
elif "qkv_proj" in r:
|
16 |
+
(q, k, v) = params[r].chunk(3)
|
17 |
+
params2[r.replace("qkv_proj", "q_proj")] = q
|
18 |
+
params2[r.replace("qkv_proj", "k_proj")] = k
|
19 |
+
params2[r.replace("qkv_proj", "v_proj")] = v
|
20 |
+
else:
|
21 |
+
params2[r] = params[r]
|
22 |
+
|
23 |
+
for r in params2.keys():
|
24 |
+
params2[r] = torch.tensor(params2[r].clone().detach(), dtype=torch.bfloat16)
|
25 |
+
|
26 |
+
save_file(params2, "model-00001-of-00001.safetensors", metadata={"format": "pt"})
|