feat(evalhub): Providers as ConfigMaps #650
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -2,3 +2,5 @@ | |
| varReference: | ||
| - kind: Deployment | ||
| path: spec/template/spec/containers[]/image | ||
| - kind: ConfigMap | ||
| path: data | ||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,7 @@ | ||
| resources: | ||
| - provider-garak.yaml | ||
| - provider-guidellm.yaml | ||
| - provider-lighteval.yaml | ||
| - provider-lm-evaluation-harness.yaml | ||
|
|
||
| namespace: system |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,80 @@ | ||
| apiVersion: v1 | ||
| kind: ConfigMap | ||
| metadata: | ||
| name: evalhub-provider-garak | ||
| labels: | ||
| trustyai.opendatahub.io/evalhub-provider-type: system | ||
| trustyai.opendatahub.io/evalhub-provider-name: garak | ||
| data: | ||
| garak.yaml: | | ||
| id: garak | ||
| name: Garak | ||
| description: LLM vulnerability scanner and red-teaming framework | ||
| type: builtin | ||
| runtime: | ||
| k8s: | ||
| image: $(evalhub-provider-garak-image) | ||
|
Member
Author
There was a problem hiding this comment. for one example, this is what the script make sure to write in manifest CM |
||
| entrypoint: | ||
| - python | ||
| - /opt/app-root/src/main.py | ||
| cpu_request: 100m | ||
| memory_request: 128Mi | ||
| cpu_limit: 500m | ||
| memory_limit: 1Gi | ||
| env: | ||
| - name: VAR_NAME | ||
| value: VALUE | ||
tarilabs marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| local: null | ||
| benchmarks: | ||
| - id: toxicity | ||
| name: Toxicity Detection | ||
| description: Tests model's tendency to generate toxic content | ||
| category: safety | ||
| metrics: | ||
| - toxicity_rate | ||
| - severity_score | ||
| num_few_shot: 0 | ||
| dataset_size: 500 | ||
| tags: | ||
| - safety | ||
| - toxicity | ||
| - red_team | ||
| - id: bias_detection | ||
| name: Bias Detection | ||
| description: Evaluates model for various forms of bias | ||
| category: fairness | ||
| metrics: | ||
| - bias_score | ||
| - demographic_parity | ||
| num_few_shot: 0 | ||
| dataset_size: 1000 | ||
| tags: | ||
| - fairness | ||
| - bias | ||
| - demographic | ||
| - id: pii_leakage | ||
| name: PII Leakage | ||
| description: Tests for personally identifiable information leakage | ||
| category: privacy | ||
| metrics: | ||
| - pii_leak_rate | ||
| - sensitivity_score | ||
| num_few_shot: 0 | ||
| dataset_size: 300 | ||
| tags: | ||
| - privacy | ||
| - pii | ||
| - security | ||
| - id: prompt_injection | ||
| name: Prompt Injection | ||
| description: Tests resilience against prompt injection attacks | ||
| category: security | ||
| metrics: | ||
| - injection_success_rate | ||
| - defense_effectiveness | ||
| num_few_shot: 0 | ||
| dataset_size: 200 | ||
| tags: | ||
| - security | ||
| - injection | ||
| - adversarial | ||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,130 @@ | ||
| apiVersion: v1 | ||
| kind: ConfigMap | ||
| metadata: | ||
| name: evalhub-provider-guidellm | ||
| labels: | ||
| trustyai.opendatahub.io/evalhub-provider-type: system | ||
| trustyai.opendatahub.io/evalhub-provider-name: guidellm | ||
| data: | ||
| guidellm.yaml: | | ||
| id: guidellm | ||
| name: GuideLLM | ||
| description: Performance benchmarking framework for LLM inference servers | ||
| type: builtin | ||
| runtime: | ||
| k8s: | ||
| image: $(evalhub-provider-guidellm-image) | ||
| entrypoint: | ||
| - python | ||
| - main.py | ||
| cpu_request: 100m | ||
| memory_request: 128Mi | ||
| cpu_limit: 1000m | ||
| memory_limit: 2Gi | ||
| local: null | ||
| benchmarks: | ||
| - id: sweep | ||
| name: Sweep Profile | ||
| description: Automatically finds optimal request rate by sweeping from low to high | ||
| concurrency | ||
| category: performance | ||
| metrics: | ||
| - requests_per_second | ||
| - prompt_tokens_per_second | ||
| - output_tokens_per_second | ||
| - mean_ttft_ms | ||
| - mean_itl_ms | ||
| tags: | ||
| - performance | ||
| - throughput | ||
| - latency | ||
| - guidellm | ||
| - auto | ||
| - id: throughput | ||
| name: Throughput Profile | ||
| description: Measures maximum throughput by gradually increasing request rate until | ||
| saturation | ||
| category: performance | ||
| metrics: | ||
| - requests_per_second | ||
| - prompt_tokens_per_second | ||
| - output_tokens_per_second | ||
| - mean_ttft_ms | ||
| - mean_itl_ms | ||
| tags: | ||
| - performance | ||
| - throughput | ||
| - guidellm | ||
| - saturation | ||
| - id: concurrent | ||
| name: Concurrent Profile | ||
| description: Tests performance with fixed number of concurrent requests | ||
| category: performance | ||
| metrics: | ||
| - requests_per_second | ||
| - prompt_tokens_per_second | ||
| - output_tokens_per_second | ||
| - mean_ttft_ms | ||
| - mean_itl_ms | ||
| tags: | ||
| - performance | ||
| - concurrency | ||
| - guidellm | ||
| - id: constant | ||
| name: Constant Rate Profile | ||
| description: Maintains constant request rate throughout benchmark duration | ||
| category: performance | ||
| metrics: | ||
| - requests_per_second | ||
| - prompt_tokens_per_second | ||
| - output_tokens_per_second | ||
| - mean_ttft_ms | ||
| - mean_itl_ms | ||
| tags: | ||
| - performance | ||
| - constant_rate | ||
| - guidellm | ||
| - id: poisson | ||
| name: Poisson Profile | ||
| description: Sends requests following Poisson distribution for realistic production | ||
| simulation | ||
| category: performance | ||
| metrics: | ||
| - requests_per_second | ||
| - prompt_tokens_per_second | ||
| - output_tokens_per_second | ||
| - mean_ttft_ms | ||
| - mean_itl_ms | ||
| tags: | ||
| - performance | ||
| - poisson | ||
| - realistic | ||
| - guidellm | ||
| - id: quick_perf_test | ||
| name: Quick Performance Test | ||
| description: Fast performance evaluation with sweep profile and limited samples | ||
| category: performance | ||
| metrics: | ||
| - requests_per_second | ||
| - mean_ttft_ms | ||
| - mean_itl_ms | ||
| tags: | ||
| - performance | ||
| - quick | ||
| - guidellm | ||
| - suite | ||
| - id: comprehensive_perf_test | ||
| name: Comprehensive Performance Test | ||
| description: Thorough performance evaluation across all profiles | ||
| category: performance | ||
| metrics: | ||
| - requests_per_second | ||
| - prompt_tokens_per_second | ||
| - output_tokens_per_second | ||
| - mean_ttft_ms | ||
| - mean_itl_ms | ||
| tags: | ||
| - performance | ||
| - comprehensive | ||
| - guidellm | ||
| - suite |
Oops, something went wrong.
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
default OOTB providers, so we don't have to force EH CR instance author to write them 🙃