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

Make llama.cpp read prompt size and seed from settings (#2299)

Browse source
This commit is contained in:
DGdev91 2023-05-25 15:29:31 +02:00 • committed by GitHub
parent ee674afa50
commit cf088566f8
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
5 changed files with 9 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

4
modules/llamacpp_model.py
View file

@ -39,8 +39,8 @@ class LlamaCppModel:
params = {
'model_path': str(path),
'n_ctx': 2048,
'seed': 0,
'n_ctx': shared.args.n_ctx,
'seed': int(shared.args.llama_cpp_seed),
'n_threads': shared.args.threads or None,
'n_batch': shared.args.n_batch,
'use_mmap': not shared.args.no_mmap,

2
modules/shared.py
View file

@ -134,6 +134,8 @@ parser.add_argument('--no-mmap', action='store_true', help='Prevent mmap from be
parser.add_argument('--mlock', action='store_true', help='Force the system to keep the model in RAM.')
parser.add_argument('--cache-capacity', type=str, help='Maximum cache capacity. Examples: 2000MiB, 2GiB. When provided without units, bytes will be assumed.')
parser.add_argument('--n-gpu-layers', type=int, default=0, help='Number of layers to offload to the GPU.')
parser.add_argument('--n_ctx', type=int, default=2048, help='Size of the prompt context.')
parser.add_argument('--llama_cpp_seed', type=int, default=0, help='Seed for llama-cpp models. Default 0 (random)')
# GPTQ
parser.add_argument('--wbits', type=int, default=0, help='Load a pre-quantized model with specified precision in bits. 2, 3, 4 and 8 are supported.')

2
modules/ui.py
View file

@ -30,7 +30,7 @@ theme = gr.themes.Default(
def list_model_elements():
elements = ['cpu_memory', 'auto_devices', 'disk', 'cpu', 'bf16', 'load_in_8bit', 'load_in_4bit', 'compute_dtype', 'quant_type', 'use_double_quant', 'wbits', 'groupsize', 'model_type', 'pre_layer', 'threads', 'n_batch', 'no_mmap', 'mlock', 'n_gpu_layers']
elements = ['cpu_memory', 'auto_devices', 'disk', 'cpu', 'bf16', 'load_in_8bit', 'load_in_4bit', 'compute_dtype', 'quant_type', 'use_double_quant', 'wbits', 'groupsize', 'model_type', 'pre_layer', 'threads', 'n_batch', 'no_mmap', 'mlock', 'n_gpu_layers', 'n_ctx', 'llama_cpp_seed']
for i in range(torch.cuda.device_count()):
elements.append(f'gpu_memory_{i}')