How to setup max_tokens for Deepseek with LM Studio?

[pstachula-dev]

I am using LM Studio with local deepseek model.
I can't increase the tokes max limit.

My configuration:

{
  "assistant": {
    "default_model": {
      "provider": "lmstudio",
      "model": "deepseek-coder-v2-lite-instruct@q4_k_m"
    },
    "version": "2"
  },
  "language_models": {
    "deepseek": {
      "api_url": "http://localhost:1234",
      "available_models": [
        {
          "provider": "deepseek",
          "name": "deepseek-coder-v2-lite-instruct@q4_k_m",
          "display_name": "Deepseek Custom",
          "max_tokens": 16000
        }
      ]
    }
  },

[berkus]

Change "language_models" section to match your provider from "assistant" section:

  "assistant": {
    "default_model": {
      "provider": "lmstudio", <-- match
      "model": "qwen2.5-coder-7b-instruct"
    },
    "version": "2"
  },
  "language_models": {
    "lmstudio": { <-- match
      "api_url": "http://localhost:1234",
      "available_models": [
        {
          "provider": "lmstudio", <-- match
          "name": "qwen2.5-coder-7b-instruct",
          "display_name": "Qwen Coder 7B Custom",
          "max_tokens": 16384
        }
      ]
    }
  },

This worked for me.