Is there an example for modifying the chats prompt template?

[IcyAlmond]

I wanted to use the chatML format but I have no clue how to configure it.

config

  {
    'gsuuon/model.nvim',
    cmd = { 'M', 'Model', 'Mchat' },
    init = function()
      vim.filetype.add({
        extension = {
          mchat = 'mchat',
        }
      })
    end,
    ft = 'mchat',

    keys = {
      {'<C-m>d', ':Mdelete<cr>', mode = 'n'},
      {'<C-m>s', ':Mselect<cr>', mode = 'n'},
      {'<C-m><space>', ':Mchat<cr>', mode = 'n' }
    },

    config = function()
      local function contents_to_strings(messages, system)
        local result = {
          '<|im_start|>' .. system .. '<|im_end|>\n',
        }


        for _, msg in ipairs(messages) do
          table.insert(result, '\n<|im_start|>' .. msg.role .. '\n' .. msg.content .. '<|im_end|>\n' )
        end

        table.insert(result, '<|im_start|>Assistant\n')

        return result
      end

      require('model').setup({
        chats = {
          Qwen = {
            provider = require('model.providers.llamacpp'),
            options = {
              model = 'Qwen2.5-Coder-7B-Instruct-Q4_K_L.gguf',
              args = {
                '-c', 8192,
                '-ngl', 99,
                '-mg', 1,
                '-ts', '0,1',
                '-fa'
              }
            },
            system = 'You are Qwen',
            create = function(input, context) return context.selection and input or '' end,
            run = function(messages, config)
              return {
                prompt = table.concat(
                  contents_to_strings(
                    messages,
                    config.system or 'config.system thing'
                  )
                ),
                '<|im_end|>\n',
                contents_to_strings = contents_to_strings,
              }
            end,
          },
        },
        prompts = {
          Qwen = {
            provider = require('model.providers.llamacpp'),
            options = {
              model = 'Qwen2.5-Coder-7B-Instruct-Q4_K_L.gguf',
              args = {
                '-c', 4096,
                '-ngl', 99,
                '-mg', 1,
                '-ts', '0,1',
                '-fa'
              }
            },
            builder = function(input, context)
              return {
                prompt =
                  '<|im_start|>'
                  .. (context.args or 'You are a helpful assistant') .. '<|im_end|>\n'
                  .. '<|im_start|>user\n'
                  .. input .. '<|im_end|>\n'
                  .. '<|im_start|>assistant\n',
                stops = { '<|im_end|>' }
              }
            end
          }
        },
      })
      require('model.providers.llamacpp').setup({
        binary = '/usr/bin/llama-server',
        models = '~/models/'
      })
    end
  },

I've got this and at least :Mchat qwen opens the menu, but trying to get it to complete just causes this:

to pop up and continue loading forever.

The format here is what i'm trying to make:

<|im_start|>system
{system_prompt}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant

[gsuuon]

Do you only see this issue with this model? I'd guess this isn't a format issue. You should get an error if the binary isn't found, but maybe some other error is causing the server to fail to start. Does launching the server separately work?

[affhp]

I get the same issue like the image you post about it displaying the model loading it forever, but actually it was loaded (I checked it in nvtop)
Also there is no notification about something like llama.cpp server is starting and llama.cpp server is started like in the demo in README.md

Instead I just wait for about 10 sec to make sure the model was completely loaded and manually delete those lines (line 2 and 3 in the image), and then move the cursor back to my first question, and go back normal mode and hit :Mchat.

The experience is not good but it is kind of work. I don't know how to configure to make it work like README.md

my config:

require("model").setup({
    chats = {
        ["qwen3-2B"] = {
            create = function(input, context)
                return context.selection and input or ""
            end,
            options = {
                args = { "-c", 4096 },
                model = "ggml-org_Qwen3-VL-2B-Instruct-GGUF_Qwen3-VL-2B-Instruct-Q8_0.gguf",
            },
            provider = require("model.providers.llamacpp"),
            run = require("model.format.zephyr").chat,
            system = "You are a helpful assistant",
        },
    },
})

require("model.providers.llamacpp").setup({
    binary = "/path/to/bin/llama-server",
    models = "~/.cache/llama.cpp",
})

require("model.util.curl")._is_debugging = true