Docs
By Goal
按优化目标或业务诉求进入。
文档正文
By Goal
按优化目标或业务诉求进入。
Low Latency
| Card | Description | Link |
|---|---|---|
| LongCat-Flash | 低时延推理 | https://gitcode.com/cann/cann-recipes-infer/-/blob/master/models/longcat-flash/README.md |
| DeepSeek-R1 / Kimi-K2 | 低时延场景部署 | https://gitcode.com/cann/cann-recipes-infer/-/blob/master/models/deepseek-r1/README.md |
High Throughput
| Card | Description | Link |
|---|---|---|
| DeepSeek-V3.2-Exp | 高吞吐推理 | https://gitcode.com/cann/cann-recipes-infer/-/blob/master/models/deepseek-v3.2-exp/README.md |
| Qwen3-235B Long Seq RL | 高吞吐训练 | https://gitcode.com/cann/cann-recipes-train/-/blob/master/llm_rl/qwen3/README.md |
Long Context
| Card | Description | Link |
|---|---|---|
| Kimi-K2-Thinking (256K) | 长序列推理 | https://gitcode.com/cann/cann-recipes-infer/-/blob/master/models/kimi-k2-thinking/README.md |
| Qwen3-235B 2K+32K RL | 长序列训练 | https://gitcode.com/cann/cann-recipes-train/-/blob/master/llm_rl/qwen3/README.md |
On-Device / Edge Deployment
| Card | Description | Link |
|---|---|---|
| HarmonyOS AscendC Ops | 端侧算子实践 | https://gitcode.com/cann/cann-recipes-harmony-infer/-/blob/master/README.md |
| Ascend 310P Embodied | 端侧具身样例 | https://gitcode.com/cann/cann-recipes-embodied-intelligence/-/blob/master/README.md |