Skip to content

Commit fb52bcf

Browse files
Ar5en1cAr5en1c
committed
chore: finalize launch polish, cache policy, and preset defaults
1 parent 30e686b commit fb52bcf

21 files changed

Lines changed: 1268 additions & 33 deletions

‎.github/ISSUE_TEMPLATE/benchmark_claim.md‎

Lines changed: 41 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,41 @@
1+
---
2+
name: Benchmark claim
3+
description: Submit a performance/model claim with reproducible evidence
4+
title: "[claim] "
5+
labels: [benchmark]
6+
assignees: []
7+
---
8+
9+
## Claim
10+
11+
State one claim only. Use `[TESTED]` / `[INFERENCE]` labels.
12+
13+
## Environment
14+
15+
- Browser + version:
16+
- OS + device:
17+
- Runtime URL:
18+
19+
## Model + Config
20+
21+
- Model A:
22+
- Model B (if compare):
23+
- Prompt tokens:
24+
- Max output tokens:
25+
- Iterations:
26+
- Temperature:
27+
- Cache mode:
28+
29+
## Evidence (required)
30+
31+
- [ ] Exported JSON attached
32+
- [ ] Screenshot attached
33+
- [ ] Prompt text (or file path) attached
34+
35+
## Result Summary
36+
37+
- Decode TPS delta:
38+
- Throughput delta:
39+
- Latency delta:
40+
- TTFT delta:
41+
- Token parity:

‎.github/ISSUE_TEMPLATE/bug_report.md‎

Lines changed: 39 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,39 @@
1+
---
2+
name: Bug report
3+
about: Report a reproducible bug in WebLLM Bench
4+
title: "[bug] "
5+
labels: [bug]
6+
assignees: []
7+
---
8+
9+
## Summary
10+
11+
What broke?
12+
13+
## Environment
14+
15+
- Browser + version:
16+
- OS + device:
17+
- URL used (local/pages):
18+
- Model ID:
19+
20+
## Repro Steps
21+
22+
1.
23+
2.
24+
3.
25+
26+
## Expected vs Actual
27+
28+
- Expected:
29+
- Actual:
30+
31+
## Artifacts (required for runtime/benchmark bugs)
32+
33+
- [ ] JSON export attached
34+
- [ ] Screenshot of result/log attached
35+
- [ ] Full benchmark config included (prompt/max/iters/temp/cache)
36+
37+
## Additional context
38+
39+
Anything else useful.

‎.github/ISSUE_TEMPLATE/bug_report.yml‎

Lines changed: 82 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,82 @@
1+
name: Bug report
2+
description: Report an app/runtime issue with reproducible details.
3+
title: "[Bug] "
4+
labels: ["bug"]
5+
body:
6+
- type: markdown
7+
attributes:
8+
value: |
9+
Thanks for reporting. Include exact runtime details so we can reproduce.
10+
- type: input
11+
id: summary
12+
attributes:
13+
label: Summary
14+
placeholder: One-line description of the problem
15+
validations:
16+
required: true
17+
- type: dropdown
18+
id: area
19+
attributes:
20+
label: Area
21+
options:
22+
- Bench
23+
- Chat
24+
- Compare
25+
- Find Best
26+
- Community
27+
- Custom model import
28+
- Build/docs/scripts
29+
validations:
30+
required: true
31+
- type: input
32+
id: browser
33+
attributes:
34+
label: Browser + version
35+
placeholder: e.g., Chrome 146.0.7680.165
36+
validations:
37+
required: true
38+
- type: input
39+
id: os
40+
attributes:
41+
label: OS + hardware
42+
placeholder: e.g., macOS 15 / M1 Air
43+
validations:
44+
required: true
45+
- type: textarea
46+
id: benchmark_profile
47+
attributes:
48+
label: Benchmark/config profile
49+
description: Include promptTokens, maxTokens, iterations, cache mode, model IDs.
50+
placeholder: promptTokens=1024, maxTokens=128, iterations=10, cache=IndexedDB ...
51+
validations:
52+
required: true
53+
- type: textarea
54+
id: steps
55+
attributes:
56+
label: Steps to reproduce
57+
placeholder: |
58+
1. Open ...
59+
2. Select ...
60+
3. Click ...
61+
validations:
62+
required: true
63+
- type: textarea
64+
id: expected
65+
attributes:
66+
label: Expected behavior
67+
validations:
68+
required: true
69+
- type: textarea
70+
id: actual
71+
attributes:
72+
label: Actual behavior
73+
validations:
74+
required: true
75+
- type: textarea
76+
id: evidence
77+
attributes:
78+
label: Evidence (required for perf/runtime issues)
79+
description: Attach exported JSON and screenshot/log snippets.
80+
placeholder: webllm-bench-YYYY-MM-DDTHHMMSS.json + screenshot/log
81+
validations:
82+
required: true

‎.github/ISSUE_TEMPLATE/config.yml‎

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,8 @@
1+
blank_issues_enabled: false
2+
contact_links:
3+
- name: WebLLM Bench Live Demo
4+
url: https://ar5en1c.github.io/webllm-bench/
5+
about: Run a quick sanity check before opening a bug.
6+
- name: Device + Model Guide
7+
url: https://github.com/Ar5en1c/webllm-bench/blob/main/docs/DEVICE_MODEL_GUIDE_2026-03-30.md
8+
about: Recommended browser/model profiles and validation workflow.

‎.github/ISSUE_TEMPLATE/feature_request.yml‎

Lines changed: 50 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,50 @@
1+
name: Feature request
2+
description: Suggest an improvement to UX, benchmarking, model support, or docs.
3+
title: "[Feature] "
4+
labels: ["enhancement"]
5+
body:
6+
- type: input
7+
id: summary
8+
attributes:
9+
label: Summary
10+
placeholder: One-line feature description
11+
validations:
12+
required: true
13+
- type: dropdown
14+
id: category
15+
attributes:
16+
label: Category
17+
options:
18+
- UX / UI
19+
- Model support
20+
- Benchmarking metrics
21+
- Docs / onboarding
22+
- Build / automation
23+
- Community features
24+
validations:
25+
required: true
26+
- type: textarea
27+
id: problem
28+
attributes:
29+
label: Problem
30+
description: What is hard today for users?
31+
validations:
32+
required: true
33+
- type: textarea
34+
id: proposal
35+
attributes:
36+
label: Proposed solution
37+
description: Describe the behavior, controls, and outputs.
38+
validations:
39+
required: true
40+
- type: textarea
41+
id: alternatives
42+
attributes:
43+
label: Alternatives considered
44+
placeholder: Any workaround or different approach
45+
- type: textarea
46+
id: success
47+
attributes:
48+
label: Success criteria
49+
description: How should we measure this was successful?
50+
placeholder: e.g., fewer init failures, faster first successful run, clearer docs path

‎.github/pull_request_template.md‎

Lines changed: 24 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,24 @@
1+
## Summary
2+
3+
What changed and why?
4+
5+
## Scope
6+
7+
- [ ] UI
8+
- [ ] Runtime logic
9+
- [ ] Docs
10+
- [ ] Tooling/CI
11+
12+
## Validation
13+
14+
- [ ] `npm run test`
15+
- [ ] Manual smoke test on `bench.html`
16+
- [ ] If perf-related: attached JSON export + screenshot
17+
18+
## Risk / Rollback
19+
20+
What could break and how to revert?
21+
22+
## Evidence
23+
24+
Attach screenshots, logs, and JSON exports for benchmark-related changes.

‎.github/workflows/ci.yml‎

Lines changed: 27 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,27 @@
1+
name: CI
2+
3+
on:
4+
push:
5+
branches: [main]
6+
pull_request:
7+
8+
jobs:
9+
validate:
10+
runs-on: ubuntu-latest
11+
steps:
12+
- name: Checkout
13+
uses: actions/checkout@v4
14+
15+
- name: Setup Node
16+
uses: actions/setup-node@v4
17+
with:
18+
node-version: '20'
19+
20+
- name: Syntax check
21+
run: node --check bench.js
22+
23+
- name: Run tests
24+
run: npm run test
25+
26+
- name: Build OSS packages
27+
run: npm run oss:pack

‎README.md‎

Lines changed: 12 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,10 @@
11
# WebLLM Bench
22
Browser-native local LLM toolkit for WebLLM/MLC models.
33

4+
Maintainer:
5+
- Kuldeep Singh ([Ar5en1c](https://github.com/Ar5en1c))
6+
- Built alongside [YouxAI](https://youxai.app)
7+
48
## Live Demo (GitHub Pages)
59

610
After enabling Pages in repository settings:
@@ -14,6 +18,11 @@ python3 -m http.server 8765
1418
# open http://localhost:8765/bench.html
1519
```
1620

21+
First-run onboarding:
22+
- [docs/START_HERE.md](docs/START_HERE.md)
23+
- [docs/TROUBLESHOOTING.md](docs/TROUBLESHOOTING.md)
24+
- [docs/HUGGINGFACE_ONE_CLICK_8K.md](docs/HUGGINGFACE_ONE_CLICK_8K.md)
25+
1726
## Core Features
1827

1928
- Benchmark (single + A/B compare)
@@ -22,13 +31,16 @@ python3 -m http.server 8765
2231
- Best-model sweep
2332
- Community baseline import/export
2433
- Custom model support (URL or local files)
34+
- One-click hosted 8k preset for Qwen2.5-1.5B
2535

2636
## Launch Artifacts
2737

2838
- Release notes: [docs/RELEASE_NOTES_v1.0.0.md](docs/RELEASE_NOTES_v1.0.0.md)
2939
- Launch thread copy: [docs/LAUNCH_THREAD_COPY_v1.0.0.md](docs/LAUNCH_THREAD_COPY_v1.0.0.md)
3040
- 8k validation protocol: [docs/VALIDATION_PROTOCOL_8K.md](docs/VALIDATION_PROTOCOL_8K.md)
3141
- Batch validation report: [reports/launch_8k_batch_validation_2026-03-28.md](reports/launch_8k_batch_validation_2026-03-28.md)
42+
- Device/model guide: [docs/DEVICE_MODEL_GUIDE_2026-03-30.md](docs/DEVICE_MODEL_GUIDE_2026-03-30.md)
43+
- Pre-launch audit: [docs/PRELAUNCH_AUDIT_2026-03-30.md](docs/PRELAUNCH_AUDIT_2026-03-30.md)
3244

3345
## Repro
3446

0 commit comments

Comments
 (0)