Add Japanese blog for DeepSpeed-FastGen (microsoft#4651)

This blog adds Japanese blog for DeepSpeed-FastGen.
(also includes small fix of typos in the original blog)

---------

Co-authored-by: Conglong Li <[email protected]>

README.md

@@ -16,7 +16,7 @@
 <b> <span style="color:orange" > DeepSpeed empowers ChatGPT-like model training with a single click, offering 15x speedup over SOTA RLHF systems with unprecedented cost reduction at all scales; [learn how](https://github.com/microsoft/DeepSpeed/tree/master/blogs/deepspeed-chat)</span>.</b>
 
 * [2023/11] [DeepSpeed ZeRO-Offload++: 6x Higher Training Throughput via Collaborative CPU/GPU Twin-Flow](https://github.com/microsoft/DeepSpeed/tree/master/blogs/deepspeed-offloadpp)
-* [2023/11] [DeepSpeed-FastGen: High-throughput Text Generation for LLMs via MII and DeepSpeed-Inference](https://github.com/microsoft/DeepSpeed/tree/master/blogs/deepspeed-fastgen) [[English](https://github.com/microsoft/DeepSpeed/tree/master/blogs/deepspeed-fastgen)] [[中文](https://github.com/microsoft/DeepSpeed/tree/master/blogs/deepspeed-fastgen/chinese/README.md)]
+* [2023/11] [DeepSpeed-FastGen: High-throughput Text Generation for LLMs via MII and DeepSpeed-Inference](https://github.com/microsoft/DeepSpeed/tree/master/blogs/deepspeed-fastgen) [[English](https://github.com/microsoft/DeepSpeed/tree/master/blogs/deepspeed-fastgen)] [[中文](https://github.com/microsoft/DeepSpeed/tree/master/blogs/deepspeed-fastgen/chinese/README.md)] [[日本語](https://github.com/microsoft/DeepSpeed/tree/master/blogs/deepspeed-fastgen/japanese/README.md)]
 * [2023/10] [DeepSpeed-VisualChat: Improve Your Chat Experience with Multi-Round Multi-Image Inputs](https://github.com/microsoft/DeepSpeed/tree/master/blogs/deepspeed-visualchat/10-03-2023/README.md) [[English](https://github.com/microsoft/DeepSpeed/tree/master/blogs/deepspeed-visualchat/10-03-2023/README.md)] [[中文](https://github.com/microsoft/DeepSpeed/blob/master/blogs/deepspeed-visualchat/10-03-2023/README-Chinese.md)] [[日本語](https://github.com/microsoft/DeepSpeed/blob/master/blogs/deepspeed-visualchat/10-03-2023/README-Japanese.md)]
 * [2023/09] Announcing the DeepSpeed4Science Initiative: Enabling large-scale scientific discovery through sophisticated AI system technologies [[DeepSpeed4Science website](https://deepspeed4science.ai/)] [[Tutorials](https://www.deepspeed.ai/deepspeed4science/)] [[White paper](https://arxiv.org/abs/2310.04610)] [[Blog](https://www.microsoft.com/en-us/research/blog/announcing-the-deepspeed4science-initiative-enabling-large-scale-scientific-discovery-through-sophisticated-ai-system-technologies/)] [[中文](https://github.com/microsoft/DeepSpeed/blob/master/blogs/deepspeed4science/chinese/README.md)] [[日本語](https://github.com/microsoft/DeepSpeed/blob/master/blogs/deepspeed4science/japanese/README.md)]
 * [2023/08] [DeepSpeed ZeRO-Inference: 20x faster inference through weight quantization and KV cache offloading](https://github.com/microsoft/DeepSpeedExamples/blob/master/inference/huggingface/zero_inference/README.md)

blogs/deepspeed-fastgen/README.md

@@ -164,8 +164,8 @@ When vLLM preempts the ongoing generation of previous requests, the generation l
 
 ### D. Token Level Timing Analysis
 
-Figure 5 displays the P50, P90, and P95 latencies of the generation processes. Both vLLM and DeepSpeed-FlexGen exhibit similar P50 latencies, but vLLM demonstrates significantly higher latencies for P90 and P95.
-Regarding the P95 latencies, DeepSpeed-FlexGen achieved a reduction of 3.7 times.
+Figure 5 displays the P50, P90, and P95 latencies of the generation processes. Both vLLM and DeepSpeed-FastGen exhibit similar P50 latencies, but vLLM demonstrates significantly higher latencies for P90 and P95.
+Regarding the P95 latencies, DeepSpeed-FastGen achieved a reduction of 3.7 times.
 
 This discrepancy is due to a noticeable spike in vLLM's generation latency when it preempts the ongoing generation to process new prompts.
 In contrast, DeepSpeed-FastGen typically processes the prompt and generation for previous requests concurrently, leading to much more consistent generation latency.

blogs/deepspeed-fastgen/chinese/README.md

@@ -164,7 +164,7 @@ DeepSpeed-FastGen 利用分块 KV 缓存和动态分割融合连续批处理，
 
 ### D. 令牌级时间分析
 
-图 5 显示了生成过程的 P50、P90 和 P95 延迟。vLLM 和 DeepSpeed-FlexGen 展示了类似的 P50 延迟，但 vLLM 的 P90 和 P95 延迟显著更高。
+图 5 显示了生成过程的 P50、P90 和 P95 延迟。vLLM 和 DeepSpeed-FastGen 展示了类似的 P50 延迟，但 vLLM 的 P90 和 P95 延迟显著更高。
 
 这种差异是由于 vLLM 在抢占正在进行的生成以处理新提示时，生成延迟出现显著增加所导致的。
 相比之下，DeepSpeed-FastGen 通常会同时处理之前请求的提示和生成，导致生成延迟更加一致。