[update] tokenizer

[fix] inference_async error

[update] llm Tokenizer

[fix] Fix model name

[fix] Fix melotts model config

[update] move the main_vlm model config file

[update] Avoid using the same tokenizer service port for LLM and VLLM…

[update] Added logic to stop inference before unit exit

[update] llm add Multi-turn conversation

[update] Use dynamically allocated tokenizer service port instead

[fix] Fix python tokenizer server cannot exit

Merge branch 'dev'

[fix] model install error

Merge pull request #10 from m5stack/dev

Merge branch 'dev' of github.com:m5stack/StackFlow into dev

[fix] Fix mode_config & tokenizer_.py PATH

[fix] Fix the bug that token was truncated incorrectly

add prefill len 256 llm model

[update] SDK && add bson

[update] del tokenizer && remove deb name

[update] README

Merge branch 'dev' of github.com:m5stack/StackFlow into dev

[update] update internVL2.5-1B version

[update] updata melotts model

[update] upload tokenizer_deepseek-r1-1.5B-ax630c.py

[update] updte mode_deepseek-r1-1.5B-ax630c.json

[fix] Fix text2token bug

[update] add deepseek-r1 model

[update] add bson support

[update] README camera