🤖 Gollama: Ollama in your terminal, Your Offline AI Copilot 🦙

🤖 Gollama: Ollama in your terminal, Your Offline AI Copilot 🦙

Gollama is a delightful tool that brings Ollama, your offline conversational AI companion, directly into your terminal. It provides a fun and interactive way to generate responses from various models without needing internet connectivity. Whether you're brainstorming ideas, exploring creative writing, or just looking for inspiration, Gollama is here to assist you.

Features

Chat TUI with History: Gollama now provides a chat-like TUI experience
with a history of previous conversations. Saves previous
conversations locally using a SQLite database to continue your conversations later.
Interactive Interface: Enjoy a seamless user experience with
intuitive interface powered by Bubble Tea.
Customizable Prompts: Tailor your prompts to get precisely the responses
you need.
Multiple Models: Choose from a variety of models to generate responses
that suit your requirements.
Visual Feedback: Stay engaged with visual cues like spinners and
formatted output.
Multimodal Support: Gollama now supports multimodal models like Llava
Model Installation & Management: Easily install and manage models using
the Ollamanager library.
Directly integrated with Gollama, refer the Ollama Model
Management section for more details.

Getting Started

Prerequisites

Ollama installed on your system or a gollama API server
accessible from your machine. (Default: http://localhost:11434,
optionally can be configured using the OLLAMA_HOST environment
variable. Refer the official Ollama Go SDK docs for further
information
At least one model installed on your Ollama server. You can install models
using the ollama pull <model-name> command.
To find a list of all available models,
check the Ollama Library.
You can also use the ollama list command to list all locally installed models.

Installation

You can install Gollama using one of the following methods:

Download the latest release

Grab the latest release from the releases page and extract the archive to a location of your choice.

Install using Go

[!NOTE] Prerequisite: Go installed on your system.

You can also install Gollama using the go install command:

go install github.com/gaurav-gosain/gollama@latest

Run using Docker

You can pull the latest docker image from the GitHub Docker Container Registry and run it using the following command:

docker run --net=host -it --rm ghcr.io/gaurav-gosain/gollama:latest

Usage

Run the executable:
```
gollama
```
or
```
/path/to/gollama
```
Follow the on-screen instructions to interact with Gollama.

[!NOTE] Running Gollama with the -h flag will display the list of available flags.

Options

TUI Specific Flags

  -v, --version Prints the version of Gollama
  -m, --manage  manages the installed Ollama models (update/delete installed models)
  -i, --install installs an Ollama model (download and install a model)
  -r, --monitor Monitor the status of running Ollama models

CLI Specific Flags

--model string   Model to use for generation
--prompt string  Prompt to use for generation
--images strings Paths to the image files to attach (png/jpg/jpeg), comma separated

[!WARNING] The responses for multimodal LLMs are slower than the normal models (also depends on the size of the attached image)

Keymaps

`Pick a chat` screen

Key	Description
`↑/k`	Up
`↓/j`	Down
`→/l/pgdn`	Next page
`←/h/pgup`	Previous page
`g/home`	Go to start
`G/end`	Go to end
`enter`	Select chat
`q`	Quit
`d`	Delete chat
`ctrl+n`	New chat
`?`	Toggle extended help

Main Chat Screen

Key	Description
`ctrl+up/k`	Move view up
`ctrl+down/j`	Move view down
`ctrl+u`	Half page up
`ctrl+d`	Half page down
`ctrl+p`	Previous message
`ctrl+n`	Next message
`ctrl+y`	Copy last response
`alt+y`	Copy highlighted message
`ctrl+o`	Toggle image picker
`ctrl+x`	Remove attachment
`ctrl+h`	Toggle help
`ctrl+c`	Exit chat

[!NOTE] The ctrl+o keybinding only works if the selected model is multimodal

Modal management screens

[!NOTE] The management screens can be chained together. For example, using the flags -imr will run Ollamanager with tabs for installing, managing, and monitoring models.

The following keybindings are common to all modal management screens:

Key	Description
`?`	Toggle help menu
`↑/k`	Move up
`↓/j`	Move down
`←/h`	Move left
`→/l`	Move right
`enter`	Pick selected item
`/`	Filter/fuzzy find items
`esc`	Clear filter
`q/ctrl+c`	Quit
`n/tab`	Switch to the next tab
`p/shift+tab`	Switch to the previous tab

[!NOTE] The following keybindings are specific to the Manage Models screen/tab:

Key	Description
`u`	Update selected model
`d`	Delete selected model

Examples

TUI Chat Mode

Ollama Model Management

[!NOTE] Gollama uses the Ollamanager library to manage models. It provides a convenient way to install, update, and delete models.

https://github.com/user-attachments/assets/9e625715-5a8a-4e71-a355-89eaa298eb9b

Piped Mode

echo "Once upon a time" | gollama --model="llama3.1" --prompt="prompt goes here"

gollama --model="llama3.1" --prompt="prompt goes here" < input.txt

CLI Mode with Images

[!IMPORTANT] Not supported for all models, check if the model is multimodal

gollama --model="llava:latest" \
 --prompt="prompt goes here" \
 --images="path/to/image.png"

[!WARNING] The --model and --prompt flags are mandatory for CLI mode. The --images flag is optional.

Local Development

Run locally using Docker

You can also run Gollama locally using docker:

Clone the repository:

git clone https://github.com/Gaurav-Gosain/gollama.git

Navigate to the project directory:
```
cd gollama
```
Build the docker image:
```
docker build -t gollama .
```
Run the docker image:
```
docker run --net=host -it gollama
```

[!NOTE] The above commands build the docker image with the tag gollama. You can replace gollama with any tag of your choice.

Build from source

If you prefer to build from source, follow these steps:

Clone the repository:

git clone https://github.com/Gaurav-Gosain/gollama.git

Navigate to the project directory:
```
cd gollama
```
Build the executable:
```
go build
```

Dependencies

Gollama relies on the following third-party packages:

ollama: The official Go SDK for ollama.
ollamanager:
A Go library for installing, managing and monitoring ollama models.
bubbletea:
A library for building terminal applications using the Model-Update-View pattern.
glamour:
A markdown rendering library for the terminal.
huh:
A library for building terminal-based forms.
lipgloss:
A library for styling text output in the terminal.

Roadmap

Implement piped mode for automated usage.
Add ability to copy responses to clipboard.
GitHub Actions for automated releases.
Add support for downloading models directly from Ollama using the rest API.
Add support for extracting and copying codeblocks from the generated responses.
Add CLI options to interact with the database and perform operations like:
- Deleting chats
- Creating a new chat
- Listing chats
- Continuing a chat from the CLI