From e0e93c103bda959faa7a3f2fcb4f88cb56a2ff1e Mon Sep 17 00:00:00 2001 From: kingbri Date: Sun, 19 Nov 2023 01:10:57 -0500 Subject: [PATCH] Sample config: Uncomment all parameters This helps clarify things when users are configuring for the first time. For example, some users were putting the model name in the "model" block instead of the "model_name" field. Signed-off-by: kingbri --- config_sample.yml | 21 ++++++++++++--------- 1 file changed, 12 insertions(+), 9 deletions(-) diff --git a/config_sample.yml b/config_sample.yml index 2fb0225..9deb861 100644 --- a/config_sample.yml +++ b/config_sample.yml @@ -1,3 +1,6 @@ +# Sample YAML file for configuration. +# Comment out values as needed. Every value has a default within the application. + # Unless specified in the comments, DO NOT put these options in quotes! # You can use https://www.yamllint.com/ if you want to check your YAML formatting. @@ -14,15 +17,15 @@ network: model: # Overrides the directory to look for models (default: models) # Windows users, DO NOT put this path in quotes! This directory will be invalid otherwise. - # model_dir: your model directory path + model_dir: your model directory path # An initial model to load. Make sure the model is located in the model directory! # A model can be loaded later via the API. - # model_name: A model name + model_name: A model name # Set the following to enable speculative decoding # draft_model_dir: your model directory path to use as draft model (path is independent from model_dir) - # draft_rope_alpha: 1.0 (default: the draft model's alpha value is calculated automatically to scale to the size of the full model.) + draft_rope_alpha: 1.0 (default: the draft model's alpha value is calculated automatically to scale to the size of the full model.) # The below parameters apply only if model_name is set @@ -33,7 +36,7 @@ model: gpu_split_auto: True # An integer array of GBs of vram to split between GPUs (default: []) - # gpu_split: [20.6, 24] + gpu_split: [20.6, 24] # Rope scaling parameters (default: 1.0) rope_scale: 1.0 @@ -46,16 +49,16 @@ model: low_mem: False # Enable 8 bit cache mode for VRAM savings (slight performance hit). Possible values FP16, FP8. (default: FP16) - # cache_mode: FP16 + cache_mode: FP16 # Options for draft models (speculative decoding). This will use more VRAM! - # draft: + draft: # Overrides the directory to look for draft (default: models) - # draft_model_dir: Your draft model directory path + draft_model_dir: Your draft model directory path # An initial draft model to load. Make sure this model is located in the model directory! # A draft model can be loaded later via the API. - # draft_model_name: A model name + draft_model_name: A model name # Rope parameters for draft models (default: 1.0) - # draft_rope_alpha: 1.0 + draft_rope_alpha: 1.0