4 commits

Author	SHA1	Message	Date
Claude	402bdc2205	test: add integration tests for Classify, BatchGenerate, Info, Metrics ... Verified on RX 7800 XT (gfx1100, ROCm 7.2): - Classify: greedy single-token via max_tokens=1 - BatchGenerate: sequential multi-prompt generation - Info: GGUF metadata (gemma3, 26 layers, Q5_K_M) - Metrics: 250 tok/s decode, 3244 MiB VRAM Co-Authored-By: Virgil <virgil@lethean.io>	2026-02-24 18:52:10 +00:00
Claude	d7db2d6e95	docs: Phase 3 complete — GGUF metadata, discovery, auto context ... Integration test verifies model discovery on real GGUF files. All 9 models in /data/lem/gguf/ discovered with correct metadata. Co-Authored-By: Virgil <virgil@lethean.io>	2026-02-19 22:24:52 +00:00
Claude	a6e647c5b7	test: graceful shutdown and concurrent request integration tests ... Clear lastErr at the start of each Generate/Chat call so that Err() reflects the most recent call, not a stale cancellation from a prior one. Add two integration tests: - GracefulShutdown: cancel mid-stream then generate again on the same model, verifying the server survives cancellation. - ConcurrentRequests: three goroutines calling Generate() simultaneously, verifying no panics or deadlocks (llama-server serialises via slots). Co-Authored-By: Virgil <virgil@lethean.io>	2026-02-19 21:50:47 +00:00
Claude	0e68d71c8a	test: integration tests for full ROCm inference pipeline ... LoadModel → Generate → Chat → Close on real AMD GPU hardware. Build-tagged //go:build rocm so normal go test skips them. Co-Authored-By: Virgil <virgil@lethean.io> Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>	2026-02-19 21:15:02 +00:00