Merge pull request #25 from lightbroker/model-support-expansion

base classes
This commit is contained in:
Adam Wilson
2025-08-18 20:25:03 -06:00
committed by GitHub
16 changed files with 167 additions and 131 deletions

View File

@@ -0,0 +1,72 @@
name: 'Reusable Test Runner | RAG + CoT | Generic'
on:
workflow_call:
inputs:
batch_offset:
description: 'Starting prompt index offset'
required: true
type: number
range_name:
description: 'Human readable range name (e.g., "1-20")'
required: true
type: string
test_file_path:
description: 'Path to the test file to run'
required: true
type: string
model_display_name:
description: 'Human readable model name for workflow title'
required: true
type: string
batch_size:
description: 'Number of prompts per batch'
required: false
type: number
default: 2
jobs:
test:
runs-on: ubuntu-latest
strategy:
# Always 10 batches per workflow
matrix:
batch: [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
fail-fast: false
steps:
- name: 'checkout'
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683
- name: 'set up Python'
uses: actions/setup-python@v3
with:
python-version: '3.12'
- name: 'set up Python dependencies'
shell: bash
run: pip install -r ${{ github.workspace }}/requirements.txt
- name: 'run text generation tests - ${{ inputs.model_display_name }} - range ${{ inputs.range_name }} batch ${{ matrix.batch }}'
shell: bash
env:
PROMPT_BATCH: ${{ matrix.batch }}
BATCH_SIZE: ${{ inputs.batch_size }}
BATCH_OFFSET: ${{ inputs.batch_offset }}
run: pytest ${{ inputs.test_file_path }} -s --disable-warnings
- name: Check for changes
id: verify-changed-files
run: |
if [ -n "$(git status --porcelain)" ]; then
echo "changed=true" >> $GITHUB_OUTPUT
else
echo "changed=false" >> $GITHUB_OUTPUT
fi
- name: Commit and push changes
if: steps.verify-changed-files.outputs.changed == 'true'
run: |
git config --local user.email "42450907+lightbroker@users.noreply.github.com"
git config --local user.name "Adam Wilson"
git add .
git commit -m "Auto-generated files from ${{ inputs.model_display_name }} range ${{ inputs.range_name }} batch ${{ matrix.batch }} [skip ci]"
git push

View File

@@ -0,0 +1,11 @@
name: '#4 (1-20) | RAG + CoT | microsoft/Phi-3-mini-4k-instruct'
on:
workflow_dispatch:
jobs:
test-prompts-1-20:
uses: ./.github/workflows/test_04_malicious_prompts_rag_and_cot_microsoft_phi_3_mini4k_instruct.base.yml
with:
batch_offset: 0
range_name: "1-20"
batch_size: 2

View File

@@ -0,0 +1,11 @@
name: '#4 (21-40) | RAG + CoT | microsoft/Phi-3-mini-4k-instruct'
on:
workflow_dispatch:
jobs:
test-prompts-21-40:
uses: ./.github/workflows/test_04_malicious_prompts_rag_and_cot_microsoft_phi_3_mini4k_instruct.base.yml
with:
batch_offset: 20
range_name: "21-40"
batch_size: 2

View File

@@ -0,0 +1,11 @@
name: '#4 (41-60) | RAG + CoT | microsoft/Phi-3-mini-4k-instruct'
on:
workflow_dispatch:
jobs:
test-prompts-41-60:
uses: ./.github/workflows/test_04_malicious_prompts_rag_and_cot_microsoft_phi_3_mini4k_instruct.base.yml
with:
batch_offset: 40
range_name: "41-60"
batch_size: 2

View File

@@ -0,0 +1,11 @@
name: '#4 (61-80) | RAG + CoT | microsoft/Phi-3-mini-4k-instruct'
on:
workflow_dispatch:
jobs:
test-prompts-61-80:
uses: ./.github/workflows/test_04_malicious_prompts_rag_and_cot_microsoft_phi_3_mini4k_instruct.base.yml
with:
batch_offset: 60
range_name: "61-80"
batch_size: 2

View File

@@ -0,0 +1,11 @@
name: '#4 (81-100) | RAG + CoT | microsoft/Phi-3-mini-4k-instruct'
on:
workflow_dispatch:
jobs:
test-prompts-81-100:
uses: ./.github/workflows/test_04_malicious_prompts_rag_and_cot_microsoft_phi_3_mini4k_instruct.base.yml
with:
batch_offset: 80
range_name: "81-100"
batch_size: 2

View File

@@ -0,0 +1,27 @@
name: 'Reusable Test #4 | RAG + CoT | meta-llama/Llama-3.2-3B-Instruct'
on:
workflow_call:
inputs:
batch_offset:
description: 'Starting prompt index offset'
required: true
type: number
range_name:
description: 'Human readable range name (e.g., "1-20")'
required: true
type: string
batch_size:
description: 'Number of prompts per batch'
required: false
type: number
default: 2
jobs:
test:
uses: ./.github/workflows/test_04.abstract_base.yml
with:
batch_offset: ${{ inputs.batch_offset }}
range_name: ${{ inputs.range_name }}
batch_size: ${{ inputs.batch_size }}
test_file_path: tests/integration/test_04_malicious_prompts_rag_and_cot_meta_llama_3_2_3b_instruct.py
model_display_name: meta-llama/Llama-3.2-3B-Instruct

View File

@@ -1,42 +0,0 @@
name: 'Test #4 | RAG + CoT | meta-llama/Llama-3.2-3B-Instruct'
on:
workflow_dispatch:
jobs:
test:
runs-on: ubuntu-latest
steps:
- name: 'checkout'
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683
- name: 'set up Python'
uses: actions/setup-python@v3
with:
python-version: '3.12'
- name: 'set up Python dependencies'
shell: bash
run: pip install -r ${{ github.workspace }}/requirements.txt
- name: 'run text generation tests'
shell: bash
run: pytest tests/integration/test_04_malicious_prompts_rag_and_cot_meta_llama_3_2_3b_instruct.py -s --disable-warnings
- name: Check for changes
id: verify-changed-files
run: |
if [ -n "$(git status --porcelain)" ]; then
echo "changed=true" >> $GITHUB_OUTPUT
else
echo "changed=false" >> $GITHUB_OUTPUT
fi
- name: Commit and push changes
if: steps.verify-changed-files.outputs.changed == 'true'
run: |
git config --local user.email "42450907+lightbroker@users.noreply.github.com"
git config --local user.name "Adam Wilson"
git add .
git commit -m "Auto-generated files from workflow [skip ci]"
git push

View File

@@ -1,4 +1,4 @@
name: '#4 | RAG + CoT | microsoft/Phi-3-mini-4k-instruct (1-20)'
name: '#4 (1-20) | RAG + CoT | microsoft/Phi-3-mini-4k-instruct'
on:
workflow_dispatch:

View File

@@ -1,4 +1,4 @@
name: '#4 | RAG + CoT | microsoft/Phi-3-mini-4k-instruct (21-40)'
name: '#4 (21-40) | RAG + CoT | microsoft/Phi-3-mini-4k-instruct'
on:
workflow_dispatch:

View File

@@ -1,4 +1,4 @@
name: '#4 | RAG + CoT | microsoft/Phi-3-mini-4k-instruct (41-60)'
name: '#4 (41-60) | RAG + CoT | microsoft/Phi-3-mini-4k-instruct'
on:
workflow_dispatch:

View File

@@ -1,4 +1,4 @@
name: '#4 | RAG + CoT | microsoft/Phi-3-mini-4k-instruct (61-80)'
name: '#4 (61-80) | RAG + CoT | microsoft/Phi-3-mini-4k-instruct'
on:
workflow_dispatch:

View File

@@ -1,4 +1,4 @@
name: '#4 | RAG + CoT | microsoft/Phi-3-mini-4k-instruct (81-100)'
name: '#4 (81-100) | RAG + CoT | microsoft/Phi-3-mini-4k-instruct'
on:
workflow_dispatch:

View File

@@ -18,42 +18,10 @@ on:
jobs:
test:
runs-on: ubuntu-latest
strategy:
# Always 10 batches per workflow
matrix:
batch: [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
fail-fast: false
steps:
- name: 'checkout'
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683
- name: 'set up Python'
uses: actions/setup-python@v3
with:
python-version: '3.12'
- name: 'set up Python dependencies'
shell: bash
run: pip install -r ${{ github.workspace }}/requirements.txt
- name: 'run text generation tests - range ${{ inputs.range_name }} batch ${{ matrix.batch }}'
shell: bash
env:
PROMPT_BATCH: ${{ matrix.batch }}
BATCH_SIZE: ${{ inputs.batch_size }}
BATCH_OFFSET: ${{ inputs.batch_offset }}
run: pytest tests/integration/test_04_malicious_prompts_rag_and_cot_microsoft_phi_3_mini4k_instruct.py -s --disable-warnings
- name: Check for changes
id: verify-changed-files
run: |
if [ -n "$(git status --porcelain)" ]; then
echo "changed=true" >> $GITHUB_OUTPUT
else
echo "changed=false" >> $GITHUB_OUTPUT
fi
- name: Commit and push changes
if: steps.verify-changed-files.outputs.changed == 'true'
run: |
git config --local user.email "42450907+lightbroker@users.noreply.github.com"
git config --local user.name "Adam Wilson"
git add .
git commit -m "Auto-generated files from range ${{ inputs.range_name }} batch ${{ matrix.batch }} [skip ci]"
git push
uses: ./.github/workflows/test_04.abstract_base.yml
with:
batch_offset: ${{ inputs.batch_offset }}
range_name: ${{ inputs.range_name }}
batch_size: ${{ inputs.batch_size }}
test_file_path: tests/integration/test_04_malicious_prompts_rag_and_cot_microsoft_phi_3_mini4k_instruct.py
model_display_name: microsoft/Phi-3-mini-4k-instruct

View File

@@ -1,44 +0,0 @@
name: 'Test #4 | RAG + CoT | microsoft/Phi-3-mini-4k-instruct | Prompts 1-20'
on:
workflow_dispatch:
jobs:
test:
runs-on: ubuntu-latest
strategy:
# Matrix for batches 1-10, each testing 2 prompts (prompts 1-20 total)
matrix:
batch: [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
fail-fast: false
steps:
- name: 'checkout'
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683
- name: 'set up Python'
uses: actions/setup-python@v3
with:
python-version: '3.12'
- name: 'set up Python dependencies'
shell: bash
run: pip install -r ${{ github.workspace }}/requirements.txt
- name: 'run text generation tests - batch ${{ matrix.batch }}'
shell: bash
env:
PROMPT_BATCH: ${{ matrix.batch }}
BATCH_SIZE: 2
BATCH_OFFSET: 0 # Prompts 1-20 (offset 0)
run: pytest tests/integration/test_04_malicious_prompts_rag_and_cot_microsoft_phi_3_mini4k_instruct.py -s --disable-warnings
- name: Check for changes
id: verify-changed-files
run: |
if [ -n "$(git status --porcelain)" ]; then
echo "changed=true" >> $GITHUB_OUTPUT
else
echo "changed=false" >> $GITHUB_OUTPUT
fi
- name: Commit and push changes (batch ${{ matrix.batch }})
if: steps.verify-changed-files.outputs.changed == 'true'
run: |
git config --local user.email "42450907+lightbroker@users.noreply.github.com"
git config --local user.name "Adam Wilson"
git add .
git commit -m "Auto-generated files from workflow batch ${{ matrix.batch }} (prompts 1-20) [skip ci]"
git push

View File

@@ -22,7 +22,7 @@ class TestRunLoggingService(AbstractTestRunLoggingService):
):
self._lock = threading.Lock()
timestamp = calendar.timegm(time.gmtime())
self.log_file_path = f"./tests/logs/test_{test_id}/{str(model_id.value).replace("/", "_")}/{start}_{end}/test_{test_id}_logs_{timestamp}.json"
self.log_file_path = f"./tests/logs/test_{test_id}/{str(model_id.value).replace("/", "_")}/{start}_{end}/test_{str(test_id).lower()}_logs_{timestamp}.json"
self._ensure_log_file_exists()
def _ensure_log_file_exists(self):