DISCLAIMER: Be aware that quantised models show reduced response quality and possible hallucinations!
Available quantization formats:
q4_k_m: Uses Q6_K for half of the attention.wv and feed_forward.w2 tensors, else Q4_K
q5_k_m: Uses Q6_K for half of the attention.wv and feed_forward.w2 tensors, else Q5_K
q6_k: Uses Q8_K for all tensors
q8_0: Almost indistinguishable from float16. High resource use and slow. Not recommended for most users.
16bit: Converted to FP16 and BF16 GGUF format.
Ollama Modfile
The GGUF file can be used with Ollama. To do this, you need to import the model using the configuration defined in the Modfile. For model eg. Bielik-Minitron-7B-v3.0-Instruct.Q4_K_M.gguf (full path to model location) Modfile looks like:
FROM ./Bielik-Minitron-7B-v3.0-Instruct.Q4_K_M.gguf
TEMPLATE """<s>{{ if .System }}<|start_header_id|>system<|end_header_id|>
{{ .System }}<|eot_id|>{{ end }}{{ if .Prompt }}<|start_header_id|>user<|end_header_id|>
{{ .Prompt }}<|eot_id|>{{ end }}<|start_header_id|>assistant<|end_header_id|>
{{ .Response }}<|eot_id|>"""
PARAMETER stop "<|start_header_id|>"
PARAMETER stop "<|end_header_id|>"
PARAMETER stop "<|eot_id|>"
# Remeber to set low temperature for experimental models (1-3bits)
PARAMETER temperature 0.1
Ollama Modfile with tools:
FROM ./Bielik-Minitron-7B-v3.0-Instruct.Q8_0.gguf
TEMPLATE """{{- /* SYSTEM + TOOLS INJECTION */ -}}
{{- if or .System .Tools -}}
<|im_start|>system
{{- if .System }}
{{ .System }}
{{- end }}
{{- if .Tools }}
You are provided with tool signatures that you can use to assist with the user's query.
You do not have to use a tool if you can respond adequately without it.
Do not make assumptions about tool arguments. If required parameters are missing, ask a clarification question.
If you decide to invoke a tool, you MUST respond with ONLY valid JSON in the following format:
{"name":"<tool-name>","arguments":{...}}
Below is a list of tools you can invoke (JSON):
{{ .Tools }}
{{- end }}
<|im_end|>
{{- end }}
{{- /* MESSAGES */ -}}
{{- range $i, $_ := .Messages }}
<|im_start|>{{ .Role }}
{{ .Content }}<|im_end|>
{{- end }}
{{- /* GENERATION PROMPT */ -}}
<|im_start|>assistant"""
PARAMETER stop "<|start_header_id|>"
PARAMETER stop "<|end_header_id|>"
PARAMETER stop "<|eot_id|>"
PARAMETER temperature 0.1
GGUF is a new format introduced by the llama.cpp team on August 21st 2023.
Here is an incomplete list of clients and libraries that are known to support GGUF:
llama.cpp. The source project for GGUF. Offers a CLI and a server option.
text-generation-webui, the most widely used web UI, with many features and powerful extensions. Supports GPU acceleration.
KoboldCpp, a fully featured web UI, with GPU accel across all platforms and GPU architectures. Especially good for story telling.
GPT4All, a free and open source local running GUI, supporting Windows, Linux and macOS with full GPU accel.
LM Studio, an easy-to-use and powerful local GUI for Windows, macOS (Silicon) and Linux, with GPU acceleration
LoLLMS Web UI, a great web UI with many interesting and unique features, including a full model library for easy model selection.
Faraday.dev, an attractive and easy to use character-based chat GUI for Windows and macOS (both Silicon and Intel), with GPU acceleration.
llama-cpp-python, a Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
candle, a Rust ML framework with a focus on performance, including GPU support, and ease of use.
ctransformers, a Python library with GPU accel, LangChain support, and OpenAI-compatible AI server. Note ctransformers has not been updated in a long time and does not support many recent models.
Responsible for model quantization
Remigiusz KinasSpeakLeash - team leadership, conceptualizing, calibration data preparation, process creation and quantized model delivery.
Kuba SołtysSpeakLeash - prepared a template with tools for Ollama
