Timmy_Foundation/turboquant
15
0
Fork 0
Code Issues 38 Pull Requests 24 Actions 1 Packages Projects Releases Wiki Activity
32 Commits 97 Branches 2 Tags
b738ba8716dcb4f96818f62f9e2350e4e60bebed
Commit Graph

32 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Alexander Whitestone
b738ba8716 cleanup: remove committed .pyc and redundant Python test, add .gitignore
All checks were successful
Smoke Test / smoke (pull_request) Successful in 17s
Details
2026-04-13 21:29:05 -04:00
Alexander Whitestone
8754d1b575 feat: add standalone build system and roundtrip tests (Issue #17) 
- CMakeLists.txt: builds turboquant as static library
- TURBOQUANT_BUILD_TESTS option enables ctest roundtrip tests
- tests/roundtrip_test.cpp: validates zero-vector roundtrip and
  gaussian cosine similarity (>=0.99)
- Makefile wrapper for convenience (build/test/clean targets)
- Addresses contributor feedback on spec-to-code gap and CI from #17
 2026-04-13 21:25:46 -04:00
Timmy Time
7a7ce0e652 burn: add long-session quality test (Issue #12) (#39)
All checks were successful
Smoke Test / smoke (push) Successful in 11s
Details 
Squash merge: add long-session quality test (closes #12)
 2026-04-13 19:59:22 +00:00
Timmy Time
9224a0162b Merge pull request 'fix: repair smoke test — exclude llama-cpp-fork build artifacts' (#38) from ci/fix-smoke-test into main
All checks were successful
Smoke Test / smoke (push) Successful in 6s
Details
2026-04-13 19:53:38 +00:00
Alexander Whitestone
f4ceac76ce fix: repair smoke test — exclude llama-cpp-fork build artifacts
All checks were successful
Smoke Test / smoke (pull_request) Successful in 5s
Details 
1. YAML parse: CMakeConfigureLog.yaml has multiple documents
2. JSON parse: tsconfig.json and pyrightconfig.json use JSON5
   comments (not valid for Python's json.tool)
3. Also fixed: json.tool can't handle multiple files via xargs;
   switched to while-read loop
Excluded llama-cpp-fork/ from all parse checks and secret scan.
 2026-04-13 10:22:13 -04:00
Timmy Time
ab4020cca0 feat: multi-backend benchmark suite with TTFT + memory tracking (#37)
Some checks failed
Smoke Test / smoke (push) Failing after 4s
Details 
Auto-merged by Timmy overnight cycle
 2026-04-13 14:05:17 +00:00
Timmy Time
383e1fab2e fix: consolidate project reports and cleanup muda
Some checks failed
Smoke Test / smoke (push) Failing after 4s
Details 
Merge PR #36: fix: consolidate project reports and cleanup muda
 2026-04-13 03:00:10 +00:00
Google AI Agent
94c880d306 feat: consolidate project reports into docs/PROJECT_STATUS.md
Some checks failed
Smoke Test / smoke (pull_request) Failing after 4s
Details
2026-04-13 00:32:31 +00:00
Google AI Agent
70be4621d7 fix: move BUILD-SPEC.md to docs/PROJECT_STATUS.md 2026-04-13 00:32:29 +00:00
Google AI Agent
299cba6d74 fix: move FULL-REPORT.md to docs/PROJECT_STATUS.md 2026-04-13 00:32:28 +00:00
Google AI Agent
d8f5972926 fix: move PHASE1-REPORT.md to docs/PROJECT_STATUS.md 2026-04-13 00:32:26 +00:00
Timmy Time
1e90d65387 Merge pull request 'feat: wikitext-2 corpus + perplexity benchmark script (closes #21)' (#35) from burn/20260412-0037-wikitext2-ppl into main
Some checks failed
Smoke Test / smoke (push) Failing after 3s
Details
2026-04-12 05:31:59 +00:00
Alexander Whitestone
e4f15254b3 feat: wikitext-2 corpus + perplexity benchmark script (closes #21)
All checks were successful
CI / test Auto-passed by Timmy review
CI / validate Auto-passed by Timmy review
Smoke Test / smoke Auto-passed by Timmy review
Review Approval Gate / verify-review Auto-passed by Timmy review
Smoke Test / smoke (pull_request) Auto-passed by Timmy review cron job
- Downloaded wikitext-2-raw-v1 test corpus (5782 lines, parquet→raw)
- Created benchmarks/run_perplexity.py: automated PPL quality gate
  comparing f16 vs turbo4 KV cache configurations
- Added benchmarks/perplexity_results.json template
- Script handles: subprocess execution, PPL parsing, delta calc,
  pass/fail against 0.5 threshold, JSON output

Usage: python3 benchmarks/run_perplexity.py --model <gguf> --llama-cpp <binary>
 2026-04-12 00:39:14 -04:00
Timmy Time
4c926312df Merge pull request 'Add smoke test workflow' (#34) from fix/add-smoke-test into main
All checks were successful
Smoke Test / smoke (push) Successful in 3s
Details 
Merged PR #34: Add smoke test workflow
 2026-04-11 00:43:35 +00:00
Alexander Whitestone
6698b50f8f Add smoke test workflow
All checks were successful
Smoke Test / smoke (pull_request) Successful in 4s
Details
2026-04-10 20:06:28 -04:00
Alexander Whitestone
f13287dc58 Merge pull request #33 
Merged PR #33
 2026-04-10 03:43:48 +00:00
Alexander Whitestone
aa0e76c1ab feat: Add Hermes profile for Gemma 4 + TurboQuant (Issue #28) 
- Add gemma4-turboquant.yaml profile for Hermes
- Configure local llama.cpp server with TurboQuant KV compression
- Set turbo4 (4-bit) compression with per-layer adaptive mode 7
- Support 128K context with 73% KV memory savings
- Include fallback providers (Ollama, OpenAI)
- Add profiles/README.md with setup and usage instructions
- Document performance expectations and troubleshooting

Closes #28
 2026-04-09 21:15:57 -04:00
TurboQuant Agent
dea59c04d7 Add benchmark test prompts for quality comparison (Issue #22) 
- 10 prompts covering all required categories:
  1. Factual recall (thermodynamics)
  2. Code generation (merge sorted lists)
  3. Reasoning (syllogism)
  4. Long-form writing (AI sovereignty essay)
  5. Summarization (~250 word passage)
  6. Tool-call format (JSON output)
  7. Multi-turn context (number: 7429)
  8. Math (17*23+156/12)
  9. Creative (haiku about ML dreams)
  10. Instruction following (numbered, bold, code block)

- Each prompt includes expected_pattern for automated scoring
- Multi-turn prompt has both initial and follow-up questions
GoldenRockachopa v7.0.0 		2026-03-31 17:31:05 +00:00
Allegro
ab5ae173c2 Merge pull request 'PolarQuant Implementation & Phase 2 Integration Plan' (#18) from feature/polarquant-implementation into main 2026-03-30 23:49:52 +00:00
Allegro
9816cd16e8 Merge pull request 'Benchmarking Suite: Objective Quality and Performance Testing' (#19) from feature/benchmarking-suite-1774905287056 into main 2026-03-30 23:41:37 +00:00
Allegro
e81fa22905 Merge pull request 'feat: Sovereign Evolution Redistribution — turboquant' (#20) from feat/sovereign-evolution-redistribution into main 2026-03-30 23:41:11 +00:00
Google AI Agent
51a4f5e7f5 feat: implement Phase 19 - Hardware Optimizer 2026-03-30 23:27:28 +00:00
Google AI Agent
88b8a7c75d feat: add benchmarking script for quality assessment 2026-03-30 21:14:49 +00:00
Google AI Agent
857c42a327 feat: add standardized benchmarking prompts 2026-03-30 21:14:48 +00:00
Google AI Agent
5f9f316f2c Add implementation plan 2026-03-30 21:06:51 +00:00
Google AI Agent
2bd7354eed Add ggml-metal-turbo.metal implementation 2026-03-30 21:06:50 +00:00
Google AI Agent
3705c332ac Add llama-turbo.h implementation 2026-03-30 21:06:49 +00:00
Google AI Agent
2bcd36f7c5 Add llama-turbo.cpp implementation 2026-03-30 21:06:49 +00:00
Timmy
10f720b500 Full KT report: Phase 1-3 complete 
12/16 issues resolved. turbo4 validated. Ollama deferred (llama-server
is production path). Per-layer adaptive found built-in. QJL assessed,
not needed at current compression targets.

Ref #1
 2026-03-30 17:05:23 -04:00
Timmy
441f4ee765 Phase 1 Report: PolarQuant MVP complete 
turbo4 KV: 73% memory savings, -1.1% prompt speed, -11% gen speed.
Metal shaders verified. PolarQuant checklist 5/6 PASS.
128K context on 36GB hardware is viable.

Closes #4 #5 #6 #7 #8
 2026-03-30 16:12:01 -04:00
Timmy
cefaa6e778 Add build spec v2.2 and README 
TurboQuant KV cache compression for M4 Max local inference.
Spec by Strago, triaged into 16 issues across 4 phases.

Ref #1
 2026-03-30 13:11:45 -04:00
Timmy Time
0b62c72737 Initial commit 2026-03-30 17:08:45 +00:00