Zabaware Forums > Ultra Hal Assistant File Sharing Area
GodelPy for Hal (Local GPT for Hal, with memory and functionality)
Spitfire2600:
Hello all!
I have updated the godel.py script to include a swear filter. I know that has been an issue for some people. The model will no longer swear, at least it shouldn't. Just replace the godel.py script from the zip file and you're set.
-Spitfire2600
brians2009:
Thanks for putting this together! An offline GPT + memory and functionality sounds amazing! I've been going through the steps and get a failure when it tries to install torch:
ERROR: Could not find a version that satisfies the requirement torch (from -r C:\Program Files (x86)\Zabaware\Ultra Hal 7\Control\Godel\requirements.txt (line 7)) (from versions: none)
The only thing I did differently was to download an Intel installation file for Python (the link you shared seems to be for an AMD chip). Is there something I should look into or a workaround to get PyTorch installed or something?
Spitfire2600:
Hey brians2009,
The python link I shared is simply a compatibility for AMD, all functionality for intel remains intact with that link.
After you install python 3.7 from my link there, you should have "pip" functions in your CMD window. Simply run "pip install torch" and it will install. No need to download any other packages.
Spitfire2600!
Shastalore:
So to make a long story short... the way I see it, in order to properly run your excellent GodelPy / Local GPT for Hal, with memory and functionality, as well as a reasonable response time in conversation (much less than your 5-8 seconds), this seems to be the required minimum hardware, better known as a gaming computer:
- Windows 10 64-bit (Windows 11 64-bit??)
- 2TB m.2 solid-state drive (How much SSD memory will GodelPy / Local GPT for Hal actually need?).
- 4.75GHz multi-core CPU
- 32 GB of ddr5 RAM (and dual-channel for best speed)
- 3060 with 12 GB of dedicated video RAM
Am I correct on this? As I, as well as others here are also wondering what type of investment will it take to properly run Local GPT for Hal without performance disappointments... although the prices for such mini gaming PCs are dropping down towards the $ 400 - $ 500 range.
Just for laughs, and to become fully familiar with your installation instructions, I installed your setup on my Windows 10 64-bit laptop: 2.16GHz Intel Celeron N2840 CPU (Mobile Intel HD graphics with shared graphics memory), 8GB DDR3L Memory, and a 500GB Crucial MX500 SATA 2.5-inch Solid State Drive (6GB/second). And, of course being only an office laptop, Local GPT for Hal didn't work.
We all thank you for your efforts on this, as I find it much more valuable than running Hal 7.5 on the cloud... with Big Brother looking over our shoulder.
Honvai:
How to add manually more knowledge to this mod?
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version