"Stop generating" button only stops the printing, llama.cpp continues

[reversebias]

Not sure if this is intended behaviour, but when hitting the "Stop generating" button, while the output being displayed does stop streaming, the llama.cpp server backend keeps generating until the stop token or the token limit.

Running a standard llama.cpp server, other frontends using the same server will stop generating as expected:
./server -m ~/models/mixtral-8x7b-instruct-v0.1.Q6_K.gguf -c 32768 -ngl 128 -ts 39,61,0 -sm row --host 0.0.0.0

.env.local

HF_TOKEN=None
USE_LOCAL_WEBSEARCH=true
MODELS=`[
  {
      "name": "mixtral-instrtuct",
      "preprompt" : "",
      "chatPromptTemplate": "<s> {{#each messages}}{{#ifUser}}[INST]{{#if @first}}{{#if @root.preprompt}}{{@root.preprompt}}\n{{/if}}{{/if}} {{content}} [/INST]{{/ifUser}}{{#ifAssistant}} {{content}}</s> {{/ifAssistant}}{{/each}}",
      "parameters": {
        "temperature" : 0.2,
        "top_p" : 0.95,
        "repetition_penalty" : 1.2,
        "top_k" : 50,
        "truncate" : 24576,
        "max_new_tokens" : 512,
        "stop" : ["</s>"]
      },
      "endpoints": [
        {
         "url": "http://127.0.0.1:8080",
         "type": "llamacpp"
        }
      ]
  }
]`