FazBrowse GitHub Viewer
|
Trending
|
URL:
|
Home
Tools:
[Download Repo ZIP]
[View Raw Code]
[Original HTTPS Page]
OneCompression/example/example_auto_run.py at main · FujitsuResearch/OneCompression · GitHub
Uh oh!
There was an error while loading.
Please reload this page
.
FujitsuResearch
/
OneCompression
Public
Notifications
You must be signed in to change notification settings
Fork
19
Star
425
Code
Issues
3
Pull requests
4
Actions
Projects
Security and quality
0
Insights
Additional navigation options
Code
Issues
Pull requests
Actions
Projects
Security and quality
Insights
Expand file tree
Breadcrumbs
OneCompression
/
example
/
example_auto_run.py
Copy path
More file actions
More file actions
Latest commit
History
History
History
23 lines (15 loc) · 655 Bytes
Breadcrumbs
OneCompression
/
example
/
example_auto_run.py
Copy path
File metadata and controls
23 lines (15 loc) · 655 Bytes
Raw
Copy raw file
Download raw file
Open symbols panel
Edit and raw actions
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
"""
Example: One-liner quantization using auto_run
Performs the following steps automatically:
1. Load the model and tokenizer from Hugging Face Hub
2. Bit-allocation based on activation-aware error estimation with ILP
3. Quantize with GPTQ + QEP
4. Evaluate perplexity (wikitext-2) and zero-shot accuracy
5. Save the quantized model to disk
Copyright 2025-2026 Fujitsu Ltd.
Author: Keiji Kimura
"""
from
onecomp
import
Runner
Runner
.
auto_run
(
model_id
=
"TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T"
)
# You can specify the total VRAM
Runner
.
auto_run
(
model_id
=
"TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T"
,
total_vram_gb
=
1
)
Back
|
FazBrowse Home
|
New Git URL