mmarch/text-generation-webui
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Add Classifier Free Guidance (CFG) for Transformers/ExLlama (#3325)

Browse source
This commit is contained in:
oobabooga 2023-08-06 17:22:48 -03:00 • committed by GitHub
parent 5134878344
commit 0af10ab49b
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
17 changed files with 131 additions and 42 deletions
Download patch file Download diff file Expand all files Collapse all files

6
server.py
View file

@ -229,7 +229,7 @@ def create_model_menus():
shared.gradio['pre_layer'] = gr.Slider(label="pre_layer", minimum=0, maximum=100, value=shared.args.pre_layer[0] if shared.args.pre_layer is not None else 0)
shared.gradio['autogptq_info'] = gr.Markdown('* ExLlama_HF is recommended over AutoGPTQ for models derived from LLaMA.')
shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
shared.gradio['max_seq_len'] = gr.Slider(label='max_seq_len', minimum=2048, maximum=16384, step=256, info='Maximum sequence length.', value=shared.args.max_seq_len)
shared.gradio['max_seq_len'] = gr.Slider(label='max_seq_len', minimum=0, maximum=16384, step=256, info='Maximum sequence length.', value=shared.args.max_seq_len)
shared.gradio['compress_pos_emb'] = gr.Slider(label='compress_pos_emb', minimum=1, maximum=8, step=1, info='Positional embeddings compression factor. Should typically be set to max_seq_len / 2048.', value=shared.args.compress_pos_emb)
shared.gradio['alpha_value'] = gr.Slider(label='alpha_value', minimum=1, maximum=32, step=1, info='Positional embeddings alpha factor for NTK RoPE scaling. Scaling is not identical to embedding compression. Use either this or compress_pos_emb, not both.', value=shared.args.alpha_value)
@ -408,6 +408,8 @@ def create_settings_menus(default_preset):
with gr.Box():
with gr.Row():
with gr.Column():
shared.gradio['guidance_scale'] = gr.Slider(-0.5, 2.5, step=0.05, value=generate_params['guidance_scale'], label='guidance_scale', info='For CFG. 1.5 is a good value.')
shared.gradio['negative_prompt'] = gr.Textbox(value=shared.settings['negative_prompt'], label='Negative prompt')
shared.gradio['mirostat_mode'] = gr.Slider(0, 2, step=1, value=generate_params['mirostat_mode'], label='mirostat_mode', info='mode=1 is for llama.cpp only.')
shared.gradio['mirostat_tau'] = gr.Slider(0, 10, step=0.01, value=generate_params['mirostat_tau'], label='mirostat_tau')
shared.gradio['mirostat_eta'] = gr.Slider(0, 1, step=0.01, value=generate_params['mirostat_eta'], label='mirostat_eta')
@ -433,7 +435,7 @@ def create_settings_menus(default_preset):
shared.gradio['stream'] = gr.Checkbox(value=not shared.args.no_stream, label='Activate text streaming')
filter_by_loader.change(loaders.blacklist_samplers, filter_by_loader, gradio(loaders.list_all_samplers()), show_progress=False)
shared.gradio['preset_menu'].change(presets.load_preset_for_ui, gradio('preset_menu', 'interface_state'), gradio('interface_state', 'do_sample', 'temperature', 'top_p', 'typical_p', 'epsilon_cutoff', 'eta_cutoff', 'repetition_penalty', 'repetition_penalty_range', 'encoder_repetition_penalty', 'top_k', 'min_length', 'no_repeat_ngram_size', 'num_beams', 'penalty_alpha', 'length_penalty', 'early_stopping', 'mirostat_mode', 'mirostat_tau', 'mirostat_eta', 'tfs', 'top_a'))
shared.gradio['preset_menu'].change(presets.load_preset_for_ui, gradio('preset_menu', 'interface_state'), gradio('interface_state') + gradio(presets.presets_params()))
def create_file_saving_menus():