@aws/ml-container-creator 0.10.0 → 0.10.3
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/LICENSE-THIRD-PARTY +9304 -0
- package/bin/cli.js +2 -0
- package/config/bootstrap-e2e-stack.json +341 -0
- package/config/bootstrap-stack.json +40 -3
- package/config/parameter-schema-v2.json +5 -21
- package/config/tune-catalog.json +1781 -0
- package/infra/ci-harness/buildspec.yml +1 -0
- package/infra/ci-harness/lambda/path-prover/brain.ts +306 -0
- package/infra/ci-harness/lambda/path-prover/write-results.ts +152 -0
- package/infra/ci-harness/lib/ci-harness-stack.ts +837 -7
- package/infra/ci-harness/state-machines/path-prover.asl.json +496 -0
- package/package.json +51 -66
- package/servers/base-image-picker/index.js +121 -121
- package/servers/e2e-status/index.js +297 -0
- package/servers/e2e-status/manifest.json +14 -0
- package/servers/e2e-status/package.json +15 -0
- package/servers/endpoint-picker/LICENSE +202 -0
- package/servers/endpoint-picker/index.js +536 -0
- package/servers/endpoint-picker/manifest.json +14 -0
- package/servers/endpoint-picker/package.json +18 -0
- package/servers/hyperpod-cluster-picker/index.js +125 -125
- package/servers/instance-sizer/index.js +138 -138
- package/servers/instance-sizer/lib/instance-ranker.js +76 -76
- package/servers/instance-sizer/lib/model-resolver.js +61 -61
- package/servers/instance-sizer/lib/quota-resolver.js +113 -113
- package/servers/instance-sizer/lib/vram-estimator.js +31 -31
- package/servers/lib/bedrock-client.js +38 -38
- package/servers/lib/catalogs/model-servers.json +201 -3
- package/servers/lib/custom-validators.js +13 -13
- package/servers/lib/dynamic-resolver.js +4 -4
- package/servers/marketplace-picker/index.js +342 -0
- package/servers/marketplace-picker/manifest.json +14 -0
- package/servers/marketplace-picker/package.json +18 -0
- package/servers/model-picker/index.js +382 -382
- package/servers/region-picker/index.js +56 -56
- package/servers/workload-picker/LICENSE +202 -0
- package/servers/workload-picker/catalogs/workload-profiles.json +67 -0
- package/servers/workload-picker/index.js +171 -0
- package/servers/workload-picker/manifest.json +16 -0
- package/servers/workload-picker/package.json +16 -0
- package/src/app.js +4 -2
- package/src/lib/bootstrap-command-handler.js +579 -14
- package/src/lib/bootstrap-config.js +36 -0
- package/src/lib/bootstrap-profile-manager.js +48 -41
- package/src/lib/ci-register-helpers.js +74 -0
- package/src/lib/config-loader.js +3 -0
- package/src/lib/config-manager.js +7 -0
- package/src/lib/cuda-resolver.js +17 -8
- package/src/lib/generated/cli-options.js +315 -315
- package/src/lib/generated/parameter-matrix.js +661 -661
- package/src/lib/generated/validation-rules.js +71 -71
- package/src/lib/path-prover-brain.js +607 -0
- package/src/lib/prompts/project-prompts.js +12 -0
- package/src/lib/template-variable-resolver.js +25 -1
- package/src/lib/tune-catalog-validator.js +37 -4
- package/templates/Dockerfile +9 -0
- package/templates/code/adapter_sidecar.py +444 -0
- package/templates/code/serve +6 -0
- package/templates/code/serve.d/vllm.ejs +1 -1
- package/templates/do/.benchmark_writer.py +1476 -0
- package/templates/do/.tune_helper.py +982 -57
- package/templates/do/__pycache__/.benchmark_writer.cpython-312.pyc +0 -0
- package/templates/do/adapter +149 -0
- package/templates/do/benchmark +639 -85
- package/templates/do/config +108 -5
- package/templates/do/deploy.d/managed-inference.ejs +192 -11
- package/templates/do/optimize +106 -37
- package/templates/do/register +89 -0
- package/templates/do/test +13 -0
- package/templates/do/tune +378 -59
- package/templates/do/validate +44 -4
|
@@ -0,0 +1,16 @@
|
|
|
1
|
+
{
|
|
2
|
+
"name": "@amzn/ml-container-creator-workload-picker",
|
|
3
|
+
"version": "1.0.0",
|
|
4
|
+
"description": "MCP server that provides named benchmark workload profiles for do/benchmark.",
|
|
5
|
+
"modes": {
|
|
6
|
+
"static": true,
|
|
7
|
+
"smart": false,
|
|
8
|
+
"discover": false
|
|
9
|
+
},
|
|
10
|
+
"catalogs": {
|
|
11
|
+
"workload-profiles": "./catalogs/workload-profiles.json"
|
|
12
|
+
},
|
|
13
|
+
"tool": {
|
|
14
|
+
"name": "get_workload_profile"
|
|
15
|
+
}
|
|
16
|
+
}
|
|
@@ -0,0 +1,16 @@
|
|
|
1
|
+
{
|
|
2
|
+
"name": "@amzn/ml-container-creator-workload-picker",
|
|
3
|
+
"private": true,
|
|
4
|
+
"version": "1.0.0",
|
|
5
|
+
"description": "MCP server that provides named benchmark workload profiles for ML Container Creator. Defines token distributions, concurrency levels, and streaming modes for standardized benchmarking.",
|
|
6
|
+
"type": "module",
|
|
7
|
+
"main": "index.js",
|
|
8
|
+
"license": "Apache-2.0",
|
|
9
|
+
"scripts": {
|
|
10
|
+
"test": "node test.js"
|
|
11
|
+
},
|
|
12
|
+
"dependencies": {
|
|
13
|
+
"@modelcontextprotocol/sdk": "^1.12.0",
|
|
14
|
+
"zod": "^3.22.0"
|
|
15
|
+
}
|
|
16
|
+
}
|
package/src/app.js
CHANGED
|
@@ -16,7 +16,7 @@ import CommentGenerator from './lib/comment-generator.js';
|
|
|
16
16
|
import ConfigurationManager from './lib/configuration-manager.js';
|
|
17
17
|
import RegistryLoader from './lib/registry-loader.js';
|
|
18
18
|
import { resolvePrefixedEnvVars } from './lib/engine-prefix-resolver.js';
|
|
19
|
-
import { _ensureTemplateVariables, _validateEnvironmentVariables
|
|
19
|
+
import { _ensureTemplateVariables, _validateEnvironmentVariables } from './lib/template-variable-resolver.js';
|
|
20
20
|
import ejs from 'ejs';
|
|
21
21
|
|
|
22
22
|
const __filename = fileURLToPath(import.meta.url);
|
|
@@ -344,13 +344,15 @@ export async function writeProject(templateDir, destDir, answers, registryConfig
|
|
|
344
344
|
// Exclude do/benchmark when benchmarking is not selected
|
|
345
345
|
if (!answers.includeBenchmark) {
|
|
346
346
|
ignorePatterns.push('**/do/benchmark');
|
|
347
|
+
ignorePatterns.push('**/do/.benchmark_writer.py');
|
|
347
348
|
ignorePatterns.push('**/do/optimize');
|
|
348
349
|
}
|
|
349
350
|
|
|
350
|
-
// Exclude do/adapter
|
|
351
|
+
// Exclude do/adapter, do/adapters/, and adapter sidecar when LoRA is not enabled
|
|
351
352
|
if (!answers.enableLora) {
|
|
352
353
|
ignorePatterns.push('**/do/adapter');
|
|
353
354
|
ignorePatterns.push('**/do/adapters/**');
|
|
355
|
+
ignorePatterns.push('**/code/adapter_sidecar.py');
|
|
354
356
|
}
|
|
355
357
|
|
|
356
358
|
// Exclude tune files when framework is NOT transformers OR deploymentTarget is batch-transform
|