初始化项目,由ModelHub XC社区提供模型
Model: Weyaxi/Einstein-v3-7B Source: Original Platform
This commit is contained in:
3
data/capybara_sharegpt.json
Normal file
3
data/capybara_sharegpt.json
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:a1947d28999416a2f468d1e09654cfdfc9bab8ccd03aa184598d20f0000dd6e4
|
||||
size 76361785
|
||||
3
data/merged_all.json
Normal file
3
data/merged_all.json
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:f915219dc691de207f9b8b448f005386330035f3c7f6cc6379dd183ac7662d2f
|
||||
size 306656943
|
||||
13
data/remove_empty_output.py
Normal file
13
data/remove_empty_output.py
Normal file
@@ -0,0 +1,13 @@
|
||||
import json
|
||||
|
||||
with open('merged_all.json', 'r') as file:
|
||||
data = json.load(file)
|
||||
|
||||
print(f"Normal len: {len(data)}")
|
||||
|
||||
data = [row for row in data if row["output"] != ""]
|
||||
|
||||
print(f"After len: {len(data)}")
|
||||
|
||||
with open('merged_all.json', 'w') as file:
|
||||
json.dump(data, file, indent=1)
|
||||
3
data/synthia-v1.3_sharegpt_12500.json
Normal file
3
data/synthia-v1.3_sharegpt_12500.json
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:dbdbc7413a3c7fc65a900518f0db8627bb5ced53e1e8ee82613d09856c1b3b70
|
||||
size 30638009
|
||||
Reference in New Issue
Block a user