1
✓
Copy the installer from the USB drive
Plug in the USB drive. Copy the installer from the mac-silicon folder to your Desktop. The models stay on the USB for now, you will load them directly in step 3.
mac-silicon → LM-Studio-0.4.20-1-arm64.dmg copy to Desktop
lmstudio-models → 3thfloor leave on USB for now
2
✓
Install LM Studio
Open LM-Studio-0.4.20-1-arm64.dmg from your Desktop. Drag LM Studio to your Applications folder, then open it. macOS may ask you to approve it, click Open.
Open LM-Studio-0.4.20-1-arm64.dmg
Drag LM Studio to Applications
Open Applications → double-click LM Studio
If macOS blocks it: System Settings → Privacy & Security → Open Anyway
Onboarding: "Your First Model" screen, skip the download
Onboarding: "Advanced Settings" screen, click Continue
3
✓
Load both models
In LM Studio, click My Models in the left sidebar. At the bottom click open in Finder. Drag the 3thfloor folder from the USB drive's lmstudio-models folder into the window. Restart LM Studio.
USB Drive → lmstudio-models → drag 3thfloor folder here
Restart LM Studio
✓ Both models appear in My Models
4
✓
Try both models
Click Chat in the left sidebar. Use the model selector at the bottom to choose qwen2.5-1.5b-instruct first. Ask a testing question. Then switch to 3thfloor-tester and ask the same question.
Model selector (bottom bar) → qwen2.5-1.5b-instruct-q4_k_m baseline first
You: What is boundary value analysis?
[ general response, no testing specialization ]
Model selector (bottom bar) → 3thfloor-tester-4b-q4_k_m now the fine-tuned model
You: What is boundary value analysis?
[ testing methodology, equivalence partitions, BBST context... ]
Working. That difference is what domain training looks like. Come back at 1:00 PM with LM Studio open.
To switch models during the workshop: click the model name at the bottom and select the other model. Both are already imported.
To switch models during the workshop: click the model name at the bottom and select the other model. Both are already imported.
1
✓
Copy the installer from the USB drive
Plug in the USB drive. Copy the installer from the windows folder to your Desktop. The models stay on the USB for now, you will load them directly in step 3.
windows → LM-Studio-0.4.20-1-x64.exe copy to Desktop
lmstudio-models → 3thfloor leave on USB for now
2
✓
Install LM Studio
Run LM-Studio-0.4.20-1-x64.exe from your Desktop. Accept the defaults. LM Studio opens when the installer finishes.
Double-click LM-Studio-0.4.20-1-x64.exe
Click Install, accept defaults
Allow if Windows Defender / SmartScreen prompts
LM Studio opens automatically when done
Onboarding: "Your First Model" screen, skip the download
Onboarding: "Advanced Settings" screen, click Continue
3
✓
Load both models
In LM Studio, click My Models in the left sidebar. At the bottom click open in file explorer. Drag the 3thfloor folder from the USB drive's lmstudio-models folder into the window. Restart LM Studio.
USB Drive → lmstudio-models → drag 3thfloor folder here
Restart LM Studio
✓ Both models appear in My Models
4
✓
Try both models
Click Chat in the left sidebar. Use the model selector at the bottom to choose qwen2.5-1.5b-instruct first. Ask a testing question. Then switch to 3thfloor-tester and ask the same question.
Model selector (bottom bar) → qwen2.5-1.5b-instruct-q4_k_m baseline first
You: What is boundary value analysis?
[ general response, no testing specialization ]
Model selector (bottom bar) → 3thfloor-tester-4b-q4_k_m now the fine-tuned model
You: What is boundary value analysis?
[ testing methodology, equivalence partitions, BBST context... ]
Working. That difference is what domain training looks like. Come back at 1:00 PM with LM Studio open.
To switch models during the workshop: click the model name at the bottom and select the other model. Both are already imported.
To switch models during the workshop: click the model name at the bottom and select the other model. Both are already imported.
1
✓
Install Ollama
Intel Mac uses Ollama instead of LM Studio, same models, terminal interface. Open the mac-intel folder on the USB drive. Unzip Ollama-darwin.zip, it extracts Ollama.app. Drag Ollama.app to your Applications folder and open it. A llama icon appears in the menu bar.
Unzip Ollama-darwin.zip → Ollama.app appears
Drag Ollama.app to Applications
Open Ollama from Applications
✓ Llama icon appears in menu bar
2
✓
Import the models
Open Terminal (Command + Space, type Terminal, press Enter). Type cd and a space, then drag the mac-intel folder from the USB into the Terminal window, it fills in the path. Press Enter. Then run the setup script.
%cd /Volumes/3th\ Floor/mac-intel
%bash setup.sh
Importing models into Ollama. This takes 1-2 minutes.
Step 1 of 2: importing baseline model (Qwen 2.5 1.5B)...
Step 2 of 2: importing fine-tuned model (3th Floor Tester 4B)...
Done.
Click the Ollama icon in your menu bar to open the chat.
Select a model from the dropdown to get started.
3
✓
Try the baseline model first
Close Terminal. Click the llama icon in your menu bar and select New Chat. Click the model dropdown and select qwen2.5:1.5b. Type your question and press Enter. This is the untrained model.
Model dropdown → qwen2.5:1.5b baseline first
You: What is boundary value analysis?
[ general response, no testing specialization ]
4
✓
Switch to the fine-tuned tester model
Click the model dropdown and switch to 3thfloor-tester. Ask the same question. On Intel, give it 15–30 seconds, that is normal.
Model dropdown → 3thfloor-tester
You: What is boundary value analysis?
[ testing methodology, equivalence partitions, BBST context, heuristics... ]
Working. That difference is what domain training looks like. Come back at 1:00 PM with the Ollama app running in your menu bar.
To switch models during the workshop: use the model dropdown in the Ollama chat window.
To switch models during the workshop: use the model dropdown in the Ollama chat window.