Update README.md

[AIwork4me]

Summary

Platform and scope

• GPU / gfx target:
• ROCm / kernel:
• Inference path: vLLM BF16 / llama.cpp GGUF / documentation / CI
• Model revision:
• vLLM or llama.cpp commit:

Verification

Evidence and claims

• New hardware claims include the hardware-validation evidence bundle.
• Benchmark changes preserve raw historical data and explain methodology impact.
• Failures and negative results are retained.
• No unverified support, performance, hash, or upstream-status claim was added.
• Relevant manifests and documentation were updated.
• No model weights, tokens, private logs, or personal data are included.