merge: resolve conflicts with SharpAI/SwiftLM main (DFlash integration)#1

Merged

ericjlake merged 72 commits into

ericjlake:perf/combinedfrom

SharpAI:merge/eric-pr85-resolved

Apr 26, 2026

Commits on Apr 21, 2026

feat: add initial dflash implementation
0xClandestine
committed

Commits on Apr 22, 2026

fix(dflash): load hiddenNorm weight + streaming + prefetch + asyncEval
0xClandestine
committed
feat: selective safetensors loader — skip expert weight data with SSD streaming
0xClandestine
committed
feat: add timings (tok/s, token count, duration) to all API responses
0xClandestine
committed
feat: add bench_35b.sh benchmark script
0xClandestine
committed
feat: add Qwen3Next SSD streaming + DFlash support
0xClandestine
committed
feat: Gemma-4 QuantizedKVCache fix + Test 9 regression (mlx-swift-lm b440)
github-actions[bot]
committed
docs+fix: kv_bits README docs + address Copilot review on PR #73
github-actions[bot]
committed
docs: expand Supported Models section to full architecture list
github-actions[bot]
committed
docs: remove GLM 5.1 from supported models (still on feature branch, reverted from main in 50c3732)
github-actions[bot]
committed
Merge pull request #73 from SharpAI/fix/gemma4-quantizedkv-b440
solderzzc
authored
make OpenAI streaming strict by default
jankaderabek
committed
test: implement OpenCode e2e validation testing with gemma-4
github-actions[bot]
committed
fix: address Copilot review — defer heartbeat cleanup, tighten tests, fix CORS/parallel test gaps
github-actions[bot]
committed
fix(ci): guard grep/jq with || true to prevent set -e abort on no-match in Tests 32-33
github-actions[bot]
committed

Commits on Apr 23, 2026

Commits on Apr 24, 2026

Commits on Apr 26, 2026

fix(server): prompt-cache bleed fixes + perf table (#85)

github-actions[bot]
and
ericjlake
committed