Spaces:
Sleeping
Sleeping
Commit History
-
Sync space files (2026-04-05)
Verified signature634ca1b
-
Rebuild: settings+new-chat visible in chat header
Verified signature4c6527e -
Fix: move Vexorium header into ChatApp, restore settings+new-chat visibility
Verified signature99c0c8d -
Fix: move Vexorium header into ChatApp, restore settings+new-chat visibility
Verified signature5598662 -
Rebuild: fix browser cache persistence
Verified signature91428e2 -
Fix cache: remove cache_dir override so env.useBrowserCache=true actually works
Verified signatureba30678 -
Rebuild: settings panel + updated examples + Vexorium style
Verified signaturef395edc -
Restore settings panel + Vexorium style: src/model-config.ts
Verified signature6f59e04 -
Restore settings panel + Vexorium style: src/components/ChatApp.tsx
Verified signature1d97bc9 -
Rebuild: header Vexorium logo, right Nemotron image
Verified signature27f3b39 -
Simplify logo layout: header Vexorium, right Nemotron only
Verified signature75352b7 -
Simplify logo layout: header Vexorium, right Nemotron only
Verified signature84520f0 -
Rebuild dist/: Nemotron × Vexorium logo
Verified signature9029782 -
Add Nemotron × Vexorium logo composition
Verified signature4709945 -
Rebuild dist/ with Vexorium brand style
Verified signaturece3dab9 -
Add Vexorium logo SVG
Verified signature87bd9b1 -
Vexorium brand: update src/components/FeatureCard.tsx
Verified signaturea87c008 -
Vexorium brand: update src/components/AppShell.tsx
Verified signatureaf63944 -
Vexorium brand: update src/components/LandingPage.tsx
Verified signature532546d -
Vexorium brand: update src/App.tsx
Verified signature0fe1132 -
Vexorium brand: update src/index.css
Verified signature0c22579 -
Rebuild dist/ with grpo-v21 model ID
Verified signature862aa2d -
Switch model to bobber/lex-interviewer-nemotron-4b-grpo-v21
Verified signatured36467f
-
add back test-webgpu.html to dist
Bobber5477c5d -
revert to 9b50a208 + modelId only, thinking enabled
Bobbera9fb427 -
deploy: fine-tuned model with thinking disabled (Q4 thinking broken on WebGPU)
BobberTest results prove the issue is model-level: - Reference model: 124 chunks, proper </think>, content works ✓ - Our Q4 model + thinking: 1 chunk, immediate <|im_end|>, zero output ✗ - Our Q4 model + no thinking: 12 chunks, good answer ✓ The Q4 quantization breaks enable_thinking=true on WebGPU. Default to thinking disabled until we fix the ONNX export.47a7f69 -
fix: adapter.requestAdapterInfo() not available in all browsers
Bobber3e3d6b1 -
fix: use explicit ESM path for transformers.js CDN import
Bobbere31aec6 -
fix: move test page to dist/ so static Space can serve it
Bobber2dcce31 -
add: WebGPU test page at /test-webgpu.html
Bobber73bc29d -
revert: back to 9b50a208 (reference model, known working)
Bobbere446d5d -
fix: handle Q4 model missing </think> on WebGPU
BobberOn WebGPU, our Q4 model never generates the </think> token, so the think-parser stays in reasoning mode and content stays empty. When this happens, move reasoning text to content so users see the response. Changes from 9b50a208 baseline: - modelId -> bobber/lex-interviewer-nemotron-4b-grpo-v12 - eos_token_id: [2, 11] (stop at <|im_end|>) - Fallback: reasoning -> content when </think> missingab0418f -
test: 9b50a208 code + only modelId changed to fine-tuned model
BobberExact same Space code that works with reference model. Only change: modelId -> bobber/lex-interviewer-nemotron-4b-grpo-v12e19a882 -
revert: back to 9b50a208 (reference model) for testing
Bobberee3637f -
fix: MessageBubble falls back to reasoning when content is empty — eliminates 'No response'
BobberThe root issue: on WebGPU, the model sometimes doesn't output </think>, so parser.content stays empty while parser.reasoning has the full response. The LLMProvider fallback tried to fix this but due to React batching, the component could render before the state update completed. This fix is at the component level: if content is empty but reasoning exists, display reasoning as the response. This is a belt-and-suspenders fix that works regardless of state timing.e08bb06 -
debug: add FINAL console.log to trace parser state — must find root cause
Bobberd4c723d
-
fix: when model skips </think>, move reasoning to content and clear reasoning
BobberOn WebGPU the Q4 model never outputs </think>, so all text goes to reasoning and content stays empty. The fix moves reasoning to content and clears the reasoning display to avoid showing duplicate text.8802730 -
fix: extract last question from reasoning when model stops after </think> without content
BobberOn WebGPU, Q4 precision differences can cause the model to output </think> then immediately <|im_end|> with no answer text. The fallback extracts the last question from the reasoning block (since the model was prompted to ask questions).4de85d0 -
fix: remove cache_dir override — use default browser Cache API for ONNX persistence
Bobberdc12d71 -
fix: handle missing </think> — fallback reasoning to content, never show 'No response'
Bobber9b9b289 -
debug: add console logging to diagnose 'No response' issue
Bobber4eb33dc -
fix: explicit eos_token_id=[2,11] at generation time — works even with cached model config
Bobber9446127 -
fix: revert to reference streamer approach + fix eos_token_id in model config
Bobber8ae6647 -
fix: filter special tokens and trailing whitespace from stream output
Bobber6572f10 -
test: float-only patch — base q4 weights + fine-tuned layernorms/biases
Bobber8a810fc -
test: use reference base model to verify pipeline works
Bobberccc9aee -
fix: remove ort-web override — model ONNX now has single external data file
BobberRoot cause was model_q4.onnx was split across _data + _data_1 files. Newer ort-web only supports single-file external data. Fixed by repacking model in HF repo to merge into single model_q4.onnx_data. Space now uses default transformers 4.0.0-next.8 + vite 8 beta (matching reference space).bf422b0 -
fix: transformers.js 4.0.0-next.8 + ort-web 1.16.3 — fixes both nemotron_h and MountedFiles errors
Bobber- Keep 4.0.0-next.8 for nemotron_h model support - Override onnxruntime-web to 1.16.3 for Module.MountedFiles external data loading - Both confirmed in built bundle2e42d02 -
fix: pin ort-web to 1.16.3 to fix Module.MountedFiles error with q4 ONNX external data
BobberThe @huggingface/transformers 4.0.0-next.8 pulls ort-web 1.22.x-dev which removed the Module.MountedFiles API used for loading external .onnx_data files. This caused: Error: Module.MountedFiles is not available Fix: downgrade to @huggingface/transformers 3.5.0 and override onnxruntime-web to 1.16.3 (last version with working MountedFiles for external data loading). Also removes bundled WASM (now served from jsDelivr CDN at runtime) and uses stable vite 5.x instead of beta 8.x.3f50c26 -
Revert to dtype=q4 — model_q4.onnx now available in our repo
bobber86a2511