History

docs: promote research citations to a top-level section on Home Moves the Google Research blog link out of Quick Links and adds both arxiv papers (Leviathan 2022 ICML, Chen 2023 DeepMind) in a dedicated "The Research Behind It" block. Signals to new readers that speculative decoding is production-grade, not experimental. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

youngharold committed Apr 24, 2026

docs: MoE support page (v0.5) Adds MoE-Support.md covering expert-aware placement, GGUF defusion, profile-guided placement, and MoE benchmarking. Links it from Home.md nav. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

youngharold committed Apr 19, 2026

docs: expand Home from three to six modes Match the website and README. Adds Combined Mode, Multi-Drafter Consensus, and Quality Gate to the landing page alongside the existing three. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

youngharold committed Apr 18, 2026

fix: update akivasolutions → youngharold in Home.md

youngharold committed Apr 10, 2026

legal: qualify output claims, add academic citations - Qualify "identical" → "identical (greedy) or statistically equivalent" - Add References section with foundational speculative decoding papers (Leviathan et al. 2023, Chen et al. 2023, SSD ICLR 2026, PEARL ICLR 2025) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

youngharold committed Mar 28, 2026

docs: add auto-tune, consensus_mode, chat_template to Configuration.md (#57) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

youngharold committed Mar 28, 2026

docs: update wiki for v0.4.1 — pipelining, perf optimizations, tree speculation Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

youngharold committed Mar 28, 2026

docs: update wiki for peer agent (#31) and consensus verification (#33) - CLI-Reference: add Peer Agents doctor section - Speculative-Decoding: add consensus verification mode documentation Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

youngharold committed Mar 28, 2026

docs: update wiki for MoE VRAM warnings (#32) - CLI-Reference: add MoE VRAM to Models check section - Troubleshooting: add auto-detection note for MoE issues Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

youngharold committed Mar 28, 2026

docs: update wiki for version enforcement (#30) and family validation (#37) - CLI-Reference: add --skip-version-check, Model Families doctor section - Hardware-Setup: note version enforcement at startup - Speculative-Decoding: note auto-detection of model families Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

youngharold committed Mar 28, 2026

Add GPT-OSS 120B benchmark results and MoE troubleshooting Benchmarking: Added MoE test results (100% acceptance, 0.39x speedup) and target speed threshold table showing when speculation helps vs hurts. Troubleshooting: Added MoE RPC distribution failure section and version matching doctor check note. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

youngharold committed Feb 20, 2026

docs: backend presets, flash_attn values, ROCm multi-GPU troubleshooting - Configuration.md: add backend presets table, extra_args/env reference, flash_attn string value docs - Troubleshooting.md: add ROCm multi-GPU hangs section with fix guidance - CLI-Reference.md: document non-interactive tightwad init flags Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

youngharold committed Feb 19, 2026

docs: document Docker healthcheck and log volume in Architecture and Configuration Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

youngharold committed Feb 18, 2026

fix: replace nonexistent Qwen3-72B with Qwen3-32B across wiki Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

youngharold committed Feb 18, 2026

docs: add dashboard section to Architecture, update CLI Reference and Home Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

youngharold committed Feb 18, 2026

docs: add distribute command reference with swarm integration Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

youngharold committed Feb 18, 2026

docs: add swarm security (--token, --allowed-ips) to wiki Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

youngharold committed Feb 18, 2026

docs: add Docker, env var config, and tightwad init documentation - Home: add Getting Started quick paths (Docker + init), update page descriptions - Architecture: add Docker deployment section with diagram and design decisions - Configuration: add env var config reference table, Docker deployment guide, Mac/Docker Desktop notes, tightwad init cross-reference - CLI-Reference: add tightwad init section with full example output - Troubleshooting: add Docker issues (LAN access, config not found, port conflicts), tightwad init issues (no servers, wrong subnet, scan timeout, single server, wrong model names) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

youngharold committed Feb 18, 2026

docs: add logs command, chat inline stats, and doctor config validation - CLI-Reference: add logs command section, config validation table, update chat section with inline stats example - Configuration: add Validation and Logging sections - Speculative-Decoding: add Chat Inline Stats section - Troubleshooting: update coordinator debug to use tightwad logs, add config validation warnings section Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

youngharold committed Feb 18, 2026

docs: add tightwad doctor to CLI Reference and Troubleshooting

youngharold committed Feb 18, 2026

docs: add Qwen3 235B, 397B, Llama 70B results to Speculative Decoding page Update tested configurations table with all cloud benchmark results. Add key findings about draft model size and cloud API latency. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

youngharold committed Feb 18, 2026

docs: add cross-family benchmark results and families script docs Add Llama 70B, Qwen3 235B, Qwen3.5 397B results to Benchmarking page. New summary table comparing all cloud configs. Document benchmark_families.py usage and config system. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

youngharold committed Feb 18, 2026

docs: add Benchmarking page, update Home and Speculative Decoding New Benchmarking wiki page with full reproduction instructions for both benchmark scripts, methodology notes, and published results table. Add Llama 3.1 8B → 405B OpenRouter results to Speculative Decoding page. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

youngharold committed Feb 18, 2026

docs: add Swarm Transfer wiki page + update CLI reference New page documenting P2P model distribution: manifest generation, seeder/puller architecture, bitfield tracking, rarest-first selection. Updated CLI Reference with manifest and swarm commands. Updated Home and Architecture pages with swarm links. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

youngharold committed Feb 18, 2026

Add next test: internet GPU pooling with speculation

youngharold committed Feb 17, 2026

Add 70B benchmark results (1.86x speedup), document same-family requirement

youngharold committed Feb 17, 2026

Fix Qwen3-72B -> Qwen2.5-72B-Instruct (Qwen3 doesn't come in 72B)

youngharold committed Feb 17, 2026

Add Pipeline Parallelism page, update Architecture with Unraid coordinator

youngharold committed Feb 17, 2026

docs: add combined mode to wiki pages Add speculation-over-RPC-pool documentation to Architecture, Configuration, and Speculative-Decoding wiki pages. Includes benchmark results, config examples, and critical requirement that llamacpp backend is needed. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

youngharold committed Feb 17, 2026

docs: add coordinator RAM requirement to Hardware Setup Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

youngharold committed Feb 17, 2026