Ollama provider #784

deathbeam · 2025-02-17T09:51:41Z

deathbeam
Feb 17, 2025
Collaborator

{
    ollama = {
        prepare_input = require('CopilotChat.config.providers').copilot.prepare_input,
        prepare_output = require('CopilotChat.config.providers').copilot.prepare_output,

        get_models = function(headers)
            local response, err = require('CopilotChat.utils').curl_get('http://localhost:11434/v1/models', {
                headers = headers,
                json_response = true,
            })

            if err then
                error(err)
            end

            return vim.tbl_map(function(model)
                return {
                    id = model.id,
                    name = model.id,
                }
            end, response.body.data)
        end,

        embed = function(inputs, headers)
            local response, err = require('CopilotChat.utils').curl_post('http://localhost:11434/v1/embeddings', {
                headers = headers,
                json_request = true,
                json_response = true,
                body = {
                    input = inputs,
                    model = 'all-minilm',
                },
            })

            if err then
                error(err)
            end

            return response.body.data
        end,

        get_url = function()
            return 'http://localhost:11434/v1/chat/completions'
        end,
    }
}

So i was messing with ollama a bit with the new providers I implemented, so I figured this might be useful to other people as well:

nach00 · 2025-03-17T07:33:20Z

nach00
Mar 17, 2025

Hi, thanks for posting this. I'm trying to get this working, but rather I am using another device (gaming pc) to host Ollama/Deepseek, and I am using my MacBook with nvim to send the prompts. Below is my config:

return {
  "CopilotC-Nvim/CopilotChat.nvim",
  branch = "main",
  cmd = "CopilotChat",
  dependencies = {
    { "nvim-lua/plenary.nvim", branch = "master" }, -- for curl, log and async functions
    { "zbirenbaum/copilot.lua" },
  },
  opts = function()
    local user = vim.env.USER or "User"
    user = user:sub(1, 1):upper() .. user:sub(2)
    return {
      -- context = "buffer",
      sticky = "#buffer",
      -- model = "claude-3.7-sonnet",
      model = "ollama",
      auto_insert_mode = true,
      question_header = "  " .. user .. " ",
      answer_header = "  Copilot ",
      window = {
        width = 0.6,
      },
      providers = {
        {
          ollama = {
            prepare_input = require("CopilotChat.config.providers").copilot.prepare_input,
            prepare_output = require("CopilotChat.config.providers").copilot.prepare_output,

            get_models = function(headers)
              local response, err = require("CopilotChat.utils").curl_get("http://192.168.1.107:11434/v1/models", {
                headers = headers,
                json_response = true,
              })

              if err then
                error(err)
              end

              return vim.tbl_map(function(model)
                return {
                  id = model.id,
                  name = model.id,
                }
              end, response.body.data)
            end,

            embed = function(inputs, headers)
              local response, err = require("CopilotChat.utils").curl_post("http://192.168.1.107:11434/v1/embeddings", {
                headers = headers,
                json_request = true,
                json_response = true,
                body = {
                  input = inputs,
                  model = "all-minilm",
                },
              })

              if err then
                error(err)
              end

              return response.body.data
            end,

            get_url = function()
              return "http://192.168.1.107:11434/v1/chat/completions"
            end,
          },
        },
      },
    }
  end,
  keys = {
    { "<leader>a?", "<cmd>CopilotChatModels<cr>", desc = "CopilotChat - Select Models" },
  },
  config = function(_, opts)
    local chat = require("CopilotChat")

    vim.api.nvim_create_autocmd("BufEnter", {
      pattern = "copilot-chat",
      callback = function()
        vim.opt_local.relativenumber = false
        vim.opt_local.number = false
      end,
    })

    chat.setup(opts)
  end,
}

Most of this config is from Lazyvim copilot chat config, I just added a keymap to select models.

Upon reloading, I get no errors, but when I send a prompt I get this error:

and

[CopilotChat] [ERROR 02:31:43] /Users/jazz/.local/share/nvim/lazy/CopilotChat.nvim/lua/CopilotChat/init.lua:683: ...re/nvim/lazy/CopilotChat.nvim/lua/CopilotChat/client.lua:344: Model not found: ollama

Does this config require tho model to be loaded directly from the same device? Or is there something wrong with my config? Thanks again.

1 reply

deathbeam Mar 17, 2025
Collaborator Author

The model is not ollama that is just provider so obviously that wont work. Use :CopilotChatModels and see the choices there that have ollama in ( ) and just copy that name

nach00 · 2025-03-18T07:15:23Z

nach00
Mar 18, 2025

Hey, thanks for the quick response. Disabling the model setting gave me this error:


                     Error executing luv callback:
...share/nvim/lazy/plenary.nvim/lua/plenary/async/async.lua:18: The coroutine failed with this message: ...re/nvim/lazy/CopilotChat.nvim/lua/CopilotChat/client.lua:227: attempt to call field 'get_headers' (a nil value)
stack traceback:
	[C]: in function 'error'
	...share/nvim/lazy/plenary.nvim/lua/plenary/async/async.lua:18: in function 'callback_or_next'
	...share/nvim/lazy/plenary.nvim/lua/plenary/async/async.lua:45: in function '_user_on_exit'
	.../.local/share/nvim/lazy/plenary.nvim/lua/plenary/job.lua:241: in function '_shutdown'
	.../.local/share/nvim/lazy/plenary.nvim/lua/plenary/job.lua:48: in function <.../.local/share/nvim/lazy/plenary.nvim/lua/plenary/job.lua:39>

I redid my config below:

return {
  "CopilotC-Nvim/CopilotChat.nvim",
  branch = "main",
  cmd = "CopilotChat",
  opts = function()
    local user = vim.env.USER or "User"
    user = user:sub(1, 1):upper() .. user:sub(2)
    return {
      auto_insert_mode = true,
      question_header = "  " .. user .. " ",
      answer_header = "  Copilot ",
      window = {
        width = 0.4,
      },

      providers = {
        github_models = {
          disabled = true,
        },

        ollama = {
          prepare_input = require("CopilotChat.config.providers").copilot.prepare_input,
          prepare_output = require("CopilotChat.config.providers").copilot.prepare_output,

          get_models = function(headers)
            local response, err = require("CopilotChat.utils").curl_get("http://192.168.107:11434/v1/models", {
              headers = config.get_headers(),
              json_response = true,
            })

            if err then
              error(err)
            end

            return vim.tbl_map(function(model)
              return {
                id = model.id,
                name = model.id,
              }
            end, response.body.data)
          end,

          embed = function(inputs, headers)
            local response, err = require("CopilotChat.utils").curl_post("http://192.168.107:11434/v1/embeddings", {
              headers = headers,
              json_request = true,
              json_response = true,
              body = {
                input = inputs,
                model = "all-minilm",
              },
            })

            if err then
              error(err)
            end

            return response.body.data
          end,

          get_url = function()
            return "http://192.168.107:11434/v1/chat/completions"
          end,
        },
      },
    }
  end,
  keys = {
    { "<c-s>", "<CR>", ft = "copilot-chat", desc = "Submit Prompt", remap = true },
    { "<leader>a", "", desc = "+ai", mode = { "n", "v" } },
    {
      "<leader>aa",
      function()
        return require("CopilotChat").toggle()
      end,
      desc = "Toggle (CopilotChat)",
      mode = { "n", "v" },
    },
    {
      "<leader>ax",
      function()
        return require("CopilotChat").reset()
      end,
      desc = "Clear (CopilotChat)",
      mode = { "n", "v" },
    },
    {
      "<leader>aq",
      function()
        vim.ui.input({
          prompt = "Quick Chat: ",
        }, function(input)
          if input ~= "" then
            require("CopilotChat").ask(input)
          end
        end)
      end,
      desc = "Quick Chat (CopilotChat)",
      mode = { "n", "v" },
    },
    {
      "<leader>ap",
      function()
        require("CopilotChat").select_prompt()
      end,
      desc = "Prompt Actions (CopilotChat)",
      mode = { "n", "v" },
    },
    { "<leader>a?", "<cmd>CopilotChatModels<cr>", desc = "CopilotChat - Select Models" },
  },
  config = function(_, opts)
    local chat = require("CopilotChat")

    vim.api.nvim_create_autocmd("BufEnter", {
      pattern = "copilot-chat",
      callback = function()
        vim.opt_local.relativenumber = false
        vim.opt_local.number = false
      end,
    })

    chat.setup(opts)
  end,
}

1 reply

deathbeam Mar 19, 2025
Collaborator Author

As mentioned on discord, the issue is the config.get_headers() call, not sure how that got there

nach00 · 2025-03-21T04:56:12Z

nach00
Mar 21, 2025

Thanks, per our discussion on discord, I removed that line and restored it to its original value... Sorry. I think I was debugging with AI and it suggested that change. Anyway, I just recreated the config below:

return {
  "CopilotC-Nvim/CopilotChat.nvim",
  branch = "main",
  cmd = "CopilotChat",
  opts = function()
    local user = vim.env.USER or "User"
    user = user:sub(1, 1):upper() .. user:sub(2)
    return {
      auto_insert_mode = true,
      question_header = "  " .. user .. " ",
      answer_header = "  Copilot ",
      window = {
        width = 0.4,
      },
      providers = {
        ollama = {
          prepare_input = require("CopilotChat.config.providers").copilot.prepare_input,
          prepare_output = require("CopilotChat.config.providers").copilot.prepare_output,

          get_models = function(headers)
            local response, err = require("CopilotChat.utils").curl_get("http://192.168.1.107:11434/v1/models", {
              headers = headers,
              json_response = true,
            })

            if err then
              error(err)
            end

            return vim.tbl_map(function(model)
              return {
                id = model.id,
                name = model.id,
              }
            end, response.body.data)
          end,

          embed = function(inputs, headers)
            local response, err = require("CopilotChat.utils").curl_post("http://192.168.1.107:11434/v1/embeddings", {
              headers = headers,
              json_request = true,
              json_response = true,
              body = {
                input = inputs,
                model = "all-minilm",
              },
            })

            if err then
              error(err)
            end

            return response.body.data
          end,

          get_url = function()
            return "http://192.168.1.107:11434/v1/chat/completions"
          end,
        },
      },
    }
  end,
  keys = {
    { "<leader>a?", "<cmd>CopilotChatModels<cr>", desc = "CopilotChat - Select Models" },
  },
  config = function(_, opts)
    local chat = require("CopilotChat")

    vim.api.nvim_create_autocmd("BufEnter", {
      pattern = "copilot-chat",
      callback = function()
        vim.opt_local.relativenumber = false
        vim.opt_local.number = false
      end,
    })

    chat.setup(opts)
  end,
}

but I'm still getting the same error when running :CopilotChatModels:

                     Error executing luv callback:
...share/nvim/lazy/plenary.nvim/lua/plenary/async/async.lua:18: The coroutine failed with this message: ...re/nvim/lazy/CopilotChat.nvim/lua/CopilotChat/client.lua:227: attempt to call field 'get_headers' (a nil value)
stack traceback:
[C]: in function 'error'
...share/nvim/lazy/plenary.nvim/lua/plenary/async/async.lua:18: in function 'callback_or_next'
...share/nvim/lazy/plenary.nvim/lua/plenary/async/async.lua:45: in function '_user_on_exit'
.../.local/share/nvim/lazy/plenary.nvim/lua/plenary/job.lua:241: in function '_shutdown'
.../.local/share/nvim/lazy/plenary.nvim/lua/plenary/job.lua:48: in function <.../.local/share/nvim/lazy/plenary.nvim/lua/plenary/job.lua:39>

and when running the /fix prompt, I get an infinite spinning wheel:

0 replies

nach00 · 2025-04-02T00:51:58Z

nach00
Apr 2, 2025

Something I just noticed. Even though my first post had the model running on my PC (and ensured it was accessible by navigating to /v1/models and see it's working), I'm still getting this error when the PC is off or the IP changed. Meaning, I don't think that the nvim plugin connects or can talk to the PC. Is there a way to check if copilotchat is able to reach the PC?

I noticed this error again when I double checked my PC's IP address to see that it changed, but was still getting the same error.

0 replies

nach00 · 2025-04-02T08:32:19Z

nach00
Apr 2, 2025

Thanks, I had to run :Lazy check and update the plugin. I thought it was automatically updated.

0 replies

richtong · 2025-04-04T02:05:47Z

richtong
Apr 4, 2025

Sorry @deathbeam just one small thing is that the snippet above has braces around ollama, but if you aren't careful, then you will bury ollama too far down. I think it may be clear documentation wise, if you say that this is the minimal addition infrastructure, just to add a provider (at least with LazyVim is

It took some trial and error to realize that the minimais:l scaffold just to add providers to LazyVim is. I'm happy to add a PR or a discussion point on how installation works here in the discussion section or in the README.md

return {
	"CopilotC-Nvim/CopilotChat.nvim",
	opts = function()
		return {
			providers = {
			     -- insert provider dictionary entries here
			    ollama = { ... },
			    openrouter = { ... },
			},
		}
	end,
}

Then you add if you say the snippet to add is just the ollama is equal to, then at least for me it is clearer how to add providers

    ollama = {
        prepare_input = require('CopilotChat.config.providers').copilot.prepare_input,
        prepare_output = require('CopilotChat.config.providers').copilot.prepare_output,

        get_models = function(headers)
            local response, err = require('CopilotChat.utils').curl_get('http://localhost:11434/v1/models', {
                headers = headers,
                json_response = true,
            })

            if err then
                error(err)
            end

            return vim.tbl_map(function(model)
                return {
                    id = model.id,
                    name = model.id,
                }
            end, response.body.data)
        end,

        embed = function(inputs, headers)
            local response, err = require('CopilotChat.utils').curl_post('http://localhost:11434/v1/embeddings', {
                headers = headers,
                json_request = true,
                json_response = true,
                body = {
                    input = inputs,
                    model = 'all-minilm',
                },
            })

            if err then
                error(err)
            end

            return response.body.data
        end,

        get_url = function()
            return 'http://localhost:11434/v1/chat/completions'
        end,
    }

1 reply

richtong Apr 4, 2025

deathbeam · 2025-04-04T02:14:59Z

deathbeam
Apr 4, 2025
Collaborator Author

Issue with changing the format is 1. if i remove braces then its invalid lua so the highlighting wont be correct, 2. dont want to use lazy.nvim specific examples as people might use different plugin managers (or none at all like me). But I agree that braces around might be a bit confusing, but not sure what would be the best solution to that

0 replies

richtong · 2025-04-04T08:23:26Z

richtong
Apr 4, 2025

Ah that makes sense. Yes ur right. I'm doing a lazyvim missing manual and will include this in it as it's lazyvim specific. Sent via Superhuman iOS ( ***@***.*** )

…

On Fri, Apr 4, 2025 at 13:15, Tomas Slusny < ***@***.*** > wrote: Issue with changing the format is 1. if i remove braces then its invalid lua so the highlighting wont be correct, 2. dont want to use lazy.nvim specific examples as people might use different plugin managers (or none at all like me). But I agree that braces around might be a bit confusing, but not sure what would be the best solution to that — Reply to this email directly, view it on GitHub ( #784 (comment) ) , or unsubscribe ( https://github.com/notifications/unsubscribe-auth/AANTCR3WHV46VXDWUIIT3N32XXTLRAVCNFSM6AAAAABXI46Q3KVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTENZSGAZTOMY ). You are receiving this because you commented. Message ID: <CopilotC-Nvim/CopilotChat . nvim/repo-discussions/784/comments/12720373 @ github. com>

0 replies

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Ollama provider #784

{{title}}

{{editor}}'s edit

{{editor}}'s edit

Replies: 8 comments 3 replies

{{title}}

{{editor}}'s edit

{{editor}}'s edit

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

Select a reply

Ollama provider #784

deathbeam Feb 17, 2025 Collaborator

Replies: 8 comments · 3 replies

deathbeam Mar 17, 2025 Collaborator Author

deathbeam Mar 19, 2025 Collaborator Author

deathbeam Apr 4, 2025 Collaborator Author

deathbeam
Feb 17, 2025
Collaborator

Replies: 8 comments 3 replies

deathbeam Mar 17, 2025
Collaborator Author

deathbeam Mar 19, 2025
Collaborator Author

deathbeam
Apr 4, 2025
Collaborator Author