versun/claude-code

mirror of https://github.com/claude-code-best/claude-code.git synced 2026-06-15 12:55:51 +00:00

Files

claude-code-best 047c85fcbf fix: 修复 DeepSeek V4 reasoning_content 回传导致的 400 错误

- 扩大模型名称检测范围，匹配所有 deepseek 模型（V4、R1 等）
- 始终保留 thinking blocks 为 reasoning_content 回传给 API
- 移除有 bug 的 turn boundary 剥离逻辑

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

2026-04-24 20:33:43 +08:00

555 B

Raw Blame History

Learner Profile

Updated: 2026-04-24

Style

Learns best with: Analogies (Memory page <-> Capsule hotel), Concrete trade-offs (Latency vs Throughput)
Strength: Strong logical intuition regarding memory constraints.
Pace: Fast. Grasped PagedAttention/TP concepts quickly from first principles.

Patterns

Tends to focus on memory usage ("is it OOM?"), which is a good instinct for inference tuning.
Needs precision on API flags (e.g., --tensor-parallel-size).

Topics

vLLM Inference Optimization (10/10 concepts mastered, 2026-04-24)