Mirrors/text-generation-webui
1
0
Fork 0
mirror of https://github.com/oobabooga/text-generation-webui.git synced 2024-11-26 09:40:20 +01:00
Code Issues Packages Projects Releases Wiki Activity

Remove mutable defaults from function signature. (#1663)

Browse Source
This commit is contained in:
IJumpAround 2023-05-08 21:55:41 -04:00 committed by GitHub
parent 32ad47c898
commit 020fe7b50b
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 11 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
modules/GPTQ_loader.py
View File

@ -35,7 +35,8 @@ except ImportError:
# This function is a replacement for the load_quant function in the # This function is a replacement for the load_quant function in the
# GPTQ-for_LLaMa repository. It supports more models and branches. # GPTQ-for_LLaMa repository. It supports more models and branches.
def _load_quant(model, checkpoint, wbits, groupsize=-1, faster_kernel=False, exclude_layers=['lm_head'], kernel_switch_threshold=128, eval=True): def _load_quant(model, checkpoint, wbits, groupsize=-1, faster_kernel=False, exclude_layers=None, kernel_switch_threshold=128, eval=True):
exclude_layers = exclude_layers or ['lm_head']
def noop(*args, **kwargs): def noop(*args, **kwargs):
pass pass

6
modules/RWKV.py
View File

@ -34,15 +34,15 @@ class RWKVModel:
result.pipeline = pipeline result.pipeline = pipeline
return result return result
def generate(self, context="", token_count=20, temperature=1, top_p=1, top_k=50, repetition_penalty=None, alpha_frequency=0.1, alpha_presence=0.1, token_ban=[0], token_stop=[], callback=None): def generate(self, context="", token_count=20, temperature=1, top_p=1, top_k=50, repetition_penalty=None, alpha_frequency=0.1, alpha_presence=0.1, token_ban=None, token_stop=None, callback=None):
args = PIPELINE_ARGS( args = PIPELINE_ARGS(
temperature=temperature, temperature=temperature,
top_p=top_p, top_p=top_p,
top_k=top_k, top_k=top_k,
alpha_frequency=alpha_frequency, # Frequency Penalty (as in GPT-3) alpha_frequency=alpha_frequency, # Frequency Penalty (as in GPT-3)
alpha_presence=alpha_presence, # Presence Penalty (as in GPT-3) alpha_presence=alpha_presence, # Presence Penalty (as in GPT-3)
token_ban=token_ban, # ban the generation of some tokens token_ban=token_ban or [0], # ban the generation of some tokens
token_stop=token_stop token_stop=token_stop or []
) )
return self.pipeline.generate(context, token_count=token_count, args=args, callback=callback) return self.pipeline.generate(context, token_count=token_count, args=args, callback=callback)

4
modules/callbacks.py
View File

@ -55,12 +55,12 @@ class Iteratorize:
Adapted from: https://stackoverflow.com/a/9969000 Adapted from: https://stackoverflow.com/a/9969000
""" """
def __init__(self, func, kwargs={}, callback=None): def __init__(self, func, kwargs=None, callback=None):
self.mfunc = func self.mfunc = func
self.c_callback = callback self.c_callback = callback
self.q = Queue() self.q = Queue()
self.sentinel = object() self.sentinel = object()
self.kwargs = kwargs self.kwargs = kwargs or {}
self.stop_now = False self.stop_now = False
def _callback(val): def _callback(val):

8
modules/text_generation.py
View File

@ -142,7 +142,7 @@ def stop_everything_event():
shared.stop_everything = True shared.stop_everything = True
def generate_reply(question, state, eos_token=None, stopping_strings=[]): def generate_reply(question, state, eos_token=None, stopping_strings=None):
state = apply_extensions('state', state) state = apply_extensions('state', state)
generate_func = apply_extensions('custom_generate_reply') generate_func = apply_extensions('custom_generate_reply')
if generate_func is None: if generate_func is None:
@ -173,7 +173,7 @@ def generate_reply(question, state, eos_token=None, stopping_strings=[]):
yield formatted_outputs(reply, shared.model_name) yield formatted_outputs(reply, shared.model_name)
def generate_reply_HF(question, original_question, seed, state, eos_token=None, stopping_strings=[]): def generate_reply_HF(question, original_question, seed, state, eos_token=None, stopping_strings=None):
generate_params = {} generate_params = {}
for k in ['max_new_tokens', 'do_sample', 'temperature', 'top_p', 'typical_p', 'repetition_penalty', 'encoder_repetition_penalty', 'top_k', 'min_length', 'no_repeat_ngram_size', 'num_beams', 'penalty_alpha', 'length_penalty', 'early_stopping']: for k in ['max_new_tokens', 'do_sample', 'temperature', 'top_p', 'typical_p', 'repetition_penalty', 'encoder_repetition_penalty', 'top_k', 'min_length', 'no_repeat_ngram_size', 'num_beams', 'penalty_alpha', 'length_penalty', 'early_stopping']:
generate_params[k] = state[k] generate_params[k] = state[k]
@ -272,7 +272,7 @@ def generate_reply_HF(question, original_question, seed, state, eos_token=None,
return return
def generate_reply_custom(question, original_question, seed, state, eos_token=None, stopping_strings=[]): def generate_reply_custom(question, original_question, seed, state, eos_token=None, stopping_strings=None):
seed = set_manual_seed(state['seed']) seed = set_manual_seed(state['seed'])
generate_params = {'token_count': state['max_new_tokens']} generate_params = {'token_count': state['max_new_tokens']}
for k in ['temperature', 'top_p', 'top_k', 'repetition_penalty']: for k in ['temperature', 'top_p', 'top_k', 'repetition_penalty']:
@ -309,7 +309,7 @@ def generate_reply_custom(question, original_question, seed, state, eos_token=No
return return
def generate_reply_flexgen(question, original_question, seed, state, eos_token=None, stopping_strings=[]): def generate_reply_flexgen(question, original_question, seed, state, eos_token=None, stopping_strings=None):
generate_params = {} generate_params = {}
for k in ['max_new_tokens', 'do_sample', 'temperature']: for k in ['max_new_tokens', 'do_sample', 'temperature']:
generate_params[k] = state[k] generate_params[k] = state[k]