Merge branch 'oobabooga:main' into stt-extension

README.md

@@ -1,6 +1,6 @@
 # Text generation web UI
 
-A gradio web UI for running Large Language Models like GPT-J 6B, OPT, GALACTICA, GPT-Neo, and Pygmalion.
+A gradio web UI for running Large Language Models like GPT-J 6B, OPT, GALACTICA, LLaMA, and Pygmalion.
 
 Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.com/AUTOMATIC1111/stable-diffusion-webui) of text generation.
 
@@ -27,6 +27,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
 * [FlexGen offload](https://github.com/oobabooga/text-generation-webui/wiki/FlexGen).
 * [DeepSpeed ZeRO-3 offload](https://github.com/oobabooga/text-generation-webui/wiki/DeepSpeed).
 * Get responses via API, [with](https://github.com/oobabooga/text-generation-webui/blob/main/api-example-streaming.py) or [without](https://github.com/oobabooga/text-generation-webui/blob/main/api-example.py) streaming.
+* [Supports the LLaMA model, including 4-bit mode](https://github.com/oobabooga/text-generation-webui/wiki/LLaMA-model).
 * [Supports the RWKV model](https://github.com/oobabooga/text-generation-webui/wiki/RWKV-model).
 * Supports softprompts.
 * [Supports extensions](https://github.com/oobabooga/text-generation-webui/wiki/Extensions).
@@ -53,7 +54,7 @@ The third line assumes that you have an NVIDIA GPU.
 pip3 install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/rocm5.2
 ```
   	  
-* If you are running in CPU mode, replace the third command with this one:
+* If you are running it in CPU mode, replace the third command with this one:
 
 ```
 conda install pytorch torchvision torchaudio git -c pytorch
@@ -137,6 +138,8 @@ Optionally, you can use the following command-line flags:
 | `--cai-chat`  | Launch the web UI in chat mode with a style similar to Character.AI's. If the file `img_bot.png` or `img_bot.jpg` exists in the same folder as server.py, this image will be used as the bot's profile picture. Similarly, `img_me.png` or `img_me.jpg` will be used as your profile picture. |
 | `--cpu`       | Use the CPU to generate text.|
 | `--load-in-8bit`  | Load the model with 8-bit precision.|
+| `--load-in-4bit`  | Load the model with 4-bit precision. Currently only works with LLaMA.|
+| `--gptq-bits GPTQ_BITS`  |  Load a pre-quantized model with specified precision. 2, 3, 4 and 8 (bit) are supported. Currently only works with LLaMA. |
 | `--bf16`  | Load the model with bfloat16 precision. Requires NVIDIA Ampere GPU. |
 | `--auto-devices` | Automatically split the model across the available GPU(s) and CPU.|
 | `--disk` | If the model is too large for your GPU(s) and CPU combined, send the remaining layers to the disk. |
@@ -176,14 +179,10 @@ Check the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/System-
 
 Pull requests, suggestions, and issue reports are welcome.
 
-Before reporting a bug, make sure that you have created a conda environment and installed the dependencies exactly as in the *Installation* section above.
+Before reporting a bug, make sure that you have:
 
-These issues are known:
-
-* 8-bit doesn't work properly on Windows or older GPUs.
-* DeepSpeed doesn't work properly on Windows.
-
-For these two, please try commenting on an existing issue instead of creating a new one.
+1. Created a conda environment and installed the dependencies exactly as in the *Installation* section above.
+2. [Searched](https://github.com/oobabooga/text-generation-webui/issues) to see if an issue already exists for the issue you encountered.
 
 ## Credits
 

download-model.py

@@ -5,7 +5,9 @@ Example:
 python download-model.py facebook/opt-1.3b
 
 '''
+
 import argparse
+import base64
 import json
 import multiprocessing
 import re
@@ -93,23 +95,28 @@ facebook/opt-1.3b
 def get_download_links_from_huggingface(model, branch):
     base = "https://huggingface.co"
     page = f"/api/models/{model}/tree/{branch}?cursor="
+    cursor = b""
 
     links = []
     classifications = []
     has_pytorch = False
     has_safetensors = False
-    while page is not None:
-        content = requests.get(f"{base}{page}").content
+    while True:
+        content = requests.get(f"{base}{page}{cursor.decode()}").content
+
         dict = json.loads(content)
+        if len(dict) == 0:
+            break
 
         for i in range(len(dict)):
             fname = dict[i]['path']
 
             is_pytorch = re.match("pytorch_model.*\.bin", fname)
             is_safetensors = re.match("model.*\.safetensors", fname)
-            is_text = re.match(".*\.(txt|json)", fname)
+            is_tokenizer = re.match("tokenizer.*\.model", fname)
+            is_text = re.match(".*\.(txt|json)", fname) or is_tokenizer
 
-            if is_text or is_safetensors or is_pytorch:
+            if any((is_pytorch, is_safetensors, is_text, is_tokenizer)):
                 if is_text:
                     links.append(f"https://huggingface.co/{model}/resolve/{branch}/{fname}")
                     classifications.append('text')
@@ -123,8 +130,9 @@ def get_download_links_from_huggingface(model, branch):
                         has_pytorch = True
                         classifications.append('pytorch')
 
-        #page = dict['nextUrl']
-        page = None
+        cursor = base64.b64encode(f'{{"file_name":"{dict[-1]["path"]}"}}'.encode()) + b':50'
+        cursor = base64.b64encode(cursor)
+        cursor = cursor.replace(b'=', b'%3D')
 
     # If both pytorch and safetensors are available, download safetensors only
     if has_pytorch and has_safetensors:

extensions/llama_prompts/script.py

@@ -0,0 +1,18 @@
+import gradio as gr
+import modules.shared as shared
+import pandas as pd
+
+df = pd.read_csv("https://raw.githubusercontent.com/devbrones/llama-prompts/main/prompts/prompts.csv")
+
+def get_prompt_by_name(name):
+    if name == 'None':
+        return ''
+    else:
+        return df[df['Prompt name'] == name].iloc[0]['Prompt'].replace('\\n', '\n')
+
+def ui():
+    if not shared.args.chat or shared.args.cai_chat:
+        choices = ['None'] + list(df['Prompt name'])
+
+        prompts_menu = gr.Dropdown(value=choices[0], choices=choices, label='Prompt')
+        prompts_menu.change(get_prompt_by_name, prompts_menu, shared.gradio['textbox'])

extensions/silero_tts/script.py

@@ -1,21 +1,45 @@
+import re
+import time
 from pathlib import Path
 
 import gradio as gr
 import torch
 
+import modules.chat as chat
+import modules.shared as shared
+
 torch._C._jit_set_profiling_mode(False)
 
 params = {
     'activate': True,
-    'speaker': 'en_56',
+    'speaker': 'en_5',
     'language': 'en',
     'model_id': 'v3_en',
     'sample_rate': 48000,
     'device': 'cpu',
+    'show_text': False,
+    'autoplay': True,
+    'voice_pitch': 'medium',
+    'voice_speed': 'medium',
 }
+
 current_params = params.copy()
 voices_by_gender = ['en_99', 'en_45', 'en_18', 'en_117', 'en_49', 'en_51', 'en_68', 'en_0', 'en_26', 'en_56', 'en_74', 'en_5', 'en_38', 'en_53', 'en_21', 'en_37', 'en_107', 'en_10', 'en_82', 'en_16', 'en_41', 'en_12', 'en_67', 'en_61', 'en_14', 'en_11', 'en_39', 'en_52', 'en_24', 'en_97', 'en_28', 'en_72', 'en_94', 'en_36', 'en_4', 'en_43', 'en_88', 'en_25', 'en_65', 'en_6', 'en_44', 'en_75', 'en_91', 'en_60', 'en_109', 'en_85', 'en_101', 'en_108', 'en_50', 'en_96', 'en_64', 'en_92', 'en_76', 'en_33', 'en_116', 'en_48', 'en_98', 'en_86', 'en_62', 'en_54', 'en_95', 'en_55', 'en_111', 'en_3', 'en_83', 'en_8', 'en_47', 'en_59', 'en_1', 'en_2', 'en_7', 'en_9', 'en_13', 'en_15', 'en_17', 'en_19', 'en_20', 'en_22', 'en_23', 'en_27', 'en_29', 'en_30', 'en_31', 'en_32', 'en_34', 'en_35', 'en_40', 'en_42', 'en_46', 'en_57', 'en_58', 'en_63', 'en_66', 'en_69', 'en_70', 'en_71', 'en_73', 'en_77', 'en_78', 'en_79', 'en_80', 'en_81', 'en_84', 'en_87', 'en_89', 'en_90', 'en_93', 'en_100', 'en_102', 'en_103', 'en_104', 'en_105', 'en_106', 'en_110', 'en_112', 'en_113', 'en_114', 'en_115']
-wav_idx = 0
+voice_pitches = ['x-low', 'low', 'medium', 'high', 'x-high']
+voice_speeds = ['x-slow', 'slow', 'medium', 'fast', 'x-fast']
+last_msg_id = 0
+
+# Used for making text xml compatible, needed for voice pitch and speed control
+table = str.maketrans({
+    "<": "&lt;",
+    ">": "&gt;",
+    "&": "&amp;",
+    "'": "&apos;",
+    '"': "&quot;",
+})
+
+def xmlesc(txt):
+    return txt.translate(table)
 
 def load_model():
     model, example_text = torch.hub.load(repo_or_dir='snakers4/silero-models', model='silero_tts', language=params['language'], speaker=params['model_id'])
@@ -33,12 +57,59 @@ def remove_surrounded_chars(string):
             new_string += char
     return new_string
 
+def remove_tts_from_history():
+    suffix = '_pygmalion' if 'pygmalion' in shared.model_name.lower() else ''
+    for i, entry in enumerate(shared.history['internal']):
+        reply = entry[1]
+        reply = re.sub("(<USER>|<user>|{{user}})", shared.settings[f'name1{suffix}'], reply)
+        if shared.args.chat:
+            reply = reply.replace('\n', '<br>')
+        shared.history['visible'][i][1] = reply
+
+    if shared.args.cai_chat:
+        return chat.generate_chat_html(shared.history['visible'], shared.settings[f'name1{suffix}'], shared.settings[f'name1{suffix}'], shared.character)
+    else:
+        return shared.history['visible']
+
+def toggle_text_in_history():
+    suffix = '_pygmalion' if 'pygmalion' in shared.model_name.lower() else ''
+    audio_str='\n\n' # The '\n\n' used after </audio>
+    if shared.args.chat:
+         audio_str='<br><br>'
+
+    if params['show_text']==True:
+        #for i, entry in enumerate(shared.history['internal']):
+        for i, entry in enumerate(shared.history['visible']):
+            vis_reply = entry[1]
+            if vis_reply.startswith('<audio'):
+                reply = shared.history['internal'][i][1]
+                reply = re.sub("(<USER>|<user>|{{user}})", shared.settings[f'name1{suffix}'], reply)
+                if shared.args.chat:
+                    reply = reply.replace('\n', '<br>')
+                shared.history['visible'][i][1] = vis_reply.split(audio_str,1)[0]+audio_str+reply
+    else:
+        for i, entry in enumerate(shared.history['visible']):
+            vis_reply = entry[1]
+            if vis_reply.startswith('<audio'):
+                shared.history['visible'][i][1] = vis_reply.split(audio_str,1)[0]+audio_str
+
+    if shared.args.cai_chat:
+        return chat.generate_chat_html(shared.history['visible'], shared.settings[f'name1{suffix}'], shared.settings[f'name1{suffix}'], shared.character)
+    else:
+        return shared.history['visible']
+
 def input_modifier(string):
     """
     This function is applied to your text inputs before
     they are fed into the model.
     """
 
+    # Remove autoplay from previous chat history
+    if (shared.args.chat or shared.args.cai_chat)and len(shared.history['internal'])>0:
+        [visible_text, visible_reply] = shared.history['visible'][-1]
+        vis_rep_clean = visible_reply.replace('controls autoplay>','controls>')
+        shared.history['visible'][-1] = [visible_text, vis_rep_clean]
+
     return string
 
 def output_modifier(string):
@@ -46,7 +117,7 @@ def output_modifier(string):
     This function is applied to the model outputs.
     """
 
-    global wav_idx, model, current_params
+    global model, current_params
 
     for i in params:
         if params[i] != current_params[i]:
@@ -57,20 +128,34 @@ def output_modifier(string):
     if params['activate'] == False:
         return string
 
+    orig_string = string
     string = remove_surrounded_chars(string)
     string = string.replace('"', '')
     string = string.replace('“', '')
     string = string.replace('\n', ' ')
     string = string.strip()
 
+    silent_string = False # Used to prevent unnecessary audio file generation
     if string == '':
         string = 'empty reply, try regenerating'
+        silent_string = True
+
+    pitch = params['voice_pitch']
+    speed = params['voice_speed']
+    prosody=f'<prosody rate="{speed}" pitch="{pitch}">'
+    string = '<speak>'+prosody+xmlesc(string)+'</prosody></speak>'
 
-    output_file = Path(f'extensions/silero_tts/outputs/{wav_idx:06d}.wav')
-    model.save_wav(text=string, speaker=params['speaker'], sample_rate=int(params['sample_rate']), audio_path=str(output_file))
+    if not shared.still_streaming and not silent_string:
+        output_file = Path(f'extensions/silero_tts/outputs/{shared.character}_{int(time.time())}.wav')
+        model.save_wav(ssml_text=string, speaker=params['speaker'], sample_rate=int(params['sample_rate']), audio_path=str(output_file))
+        autoplay_str = ' autoplay' if params['autoplay'] else ''
+        string = f'<audio src="file/{output_file.as_posix()}" controls{autoplay_str}></audio>\n\n'
+    else:
+        # Placeholder so text doesn't shift around so much
+        string = '<audio controls></audio>\n\n'
 
-    string = f'<audio src="file/{output_file.as_posix()}" controls></audio>'
-    wav_idx += 1
+    if params['show_text']:
+        string += orig_string
 
     return string
 
@@ -85,9 +170,36 @@ def bot_prefix_modifier(string):
 
 def ui():
     # Gradio elements
-    activate = gr.Checkbox(value=params['activate'], label='Activate TTS')
-    voice = gr.Dropdown(value=params['speaker'], choices=voices_by_gender, label='TTS voice')
+    with gr.Accordion("Silero TTS"):
+        with gr.Row():
+            activate = gr.Checkbox(value=params['activate'], label='Activate TTS')
+            autoplay = gr.Checkbox(value=params['autoplay'], label='Play TTS automatically')
+        show_text = gr.Checkbox(value=params['show_text'], label='Show message text under audio player')
+        voice = gr.Dropdown(value=params['speaker'], choices=voices_by_gender, label='TTS voice')
+        with gr.Row():
+            v_pitch = gr.Dropdown(value=params['voice_pitch'], choices=voice_pitches, label='Voice pitch')
+            v_speed = gr.Dropdown(value=params['voice_speed'], choices=voice_speeds, label='Voice speed')
+        with gr.Row():
+            convert = gr.Button('Permanently replace chat history audio with message text')
+            convert_confirm = gr.Button('Confirm (cannot be undone)', variant="stop", visible=False)
+            convert_cancel = gr.Button('Cancel', visible=False)
+
+    # Convert history with confirmation
+    convert_arr = [convert_confirm, convert, convert_cancel]
+    convert.click(lambda :[gr.update(visible=True), gr.update(visible=False), gr.update(visible=True)], None, convert_arr)
+    convert_confirm.click(lambda :[gr.update(visible=False), gr.update(visible=True), gr.update(visible=False)], None, convert_arr)
+    convert_confirm.click(remove_tts_from_history, [], shared.gradio['display'])
+    convert_confirm.click(lambda : chat.save_history(timestamp=False), [], [], show_progress=False)
+    convert_cancel.click(lambda :[gr.update(visible=False), gr.update(visible=True), gr.update(visible=False)], None, convert_arr)
+
+    # Toggle message text in history
+    show_text.change(lambda x: params.update({"show_text": x}), show_text, None)
+    show_text.change(toggle_text_in_history, [], shared.gradio['display'])
+    show_text.change(lambda : chat.save_history(timestamp=False), [], [], show_progress=False)
 
     # Event functions to update the parameters in the backend
     activate.change(lambda x: params.update({"activate": x}), activate, None)
+    autoplay.change(lambda x: params.update({"autoplay": x}), autoplay, None)
     voice.change(lambda x: params.update({"speaker": x}), voice, None)
+    v_pitch.change(lambda x: params.update({"voice_pitch": x}), v_pitch, None)
+    v_speed.change(lambda x: params.update({"voice_speed": x}), v_speed, None)

modules/RWKV.py

@@ -1,12 +1,11 @@
 import os
 from pathlib import Path
-from queue import Queue
-from threading import Thread
 
 import numpy as np
 from tokenizers import Tokenizer
 
 import modules.shared as shared
+from modules.callbacks import Iteratorize
 
 np.set_printoptions(precision=4, suppress=True, linewidth=200)
 
@@ -49,11 +48,11 @@ class RWKVModel:
         return context+self.pipeline.generate(context, token_count=token_count, args=args, callback=callback)
 
     def generate_with_streaming(self, **kwargs):
-        iterable = Iteratorize(self.generate, kwargs, callback=None)
-        reply = kwargs['context']
-        for token in iterable:
-            reply += token
-            yield reply
+        with Iteratorize(self.generate, kwargs, callback=None) as generator:
+            reply = kwargs['context']
+            for token in generator:
+                reply += token
+                yield reply
 
 class RWKVTokenizer:
     def __init__(self):
@@ -73,38 +72,3 @@ class RWKVTokenizer:
 
     def decode(self, ids):
         return self.tokenizer.decode(ids)
-
-class Iteratorize:
-
-    """
-    Transforms a function that takes a callback
-    into a lazy iterator (generator).
-    """
-
-    def __init__(self, func, kwargs={}, callback=None):
-        self.mfunc=func
-        self.c_callback=callback
-        self.q = Queue(maxsize=1)
-        self.sentinel = object()
-        self.kwargs = kwargs
-
-        def _callback(val):
-            self.q.put(val)
-
-        def gentask():
-            ret = self.mfunc(callback=_callback, **self.kwargs)
-            self.q.put(self.sentinel)
-            if self.c_callback:
-                self.c_callback(ret)
-
-        Thread(target=gentask).start()
-
-    def __iter__(self):
-        return self
-
-    def __next__(self):
-        obj = self.q.get(True,None)
-        if obj is self.sentinel:
-            raise StopIteration
-        else:
-            return obj

modules/callbacks.py

@@ -0,0 +1,98 @@
+import gc
+from queue import Queue
+from threading import Thread
+
+import torch
+import transformers
+
+import modules.shared as shared
+
+# Copied from https://github.com/PygmalionAI/gradio-ui/
+class _SentinelTokenStoppingCriteria(transformers.StoppingCriteria):
+
+    def __init__(self, sentinel_token_ids: torch.LongTensor,
+                 starting_idx: int):
+        transformers.StoppingCriteria.__init__(self)
+        self.sentinel_token_ids = sentinel_token_ids
+        self.starting_idx = starting_idx
+
+    def __call__(self, input_ids: torch.LongTensor,
+                 _scores: torch.FloatTensor) -> bool:
+        for sample in input_ids:
+            trimmed_sample = sample[self.starting_idx:]
+            # Can't unfold, output is still too tiny. Skip.
+            if trimmed_sample.shape[-1] < self.sentinel_token_ids.shape[-1]:
+                continue
+
+            for window in trimmed_sample.unfold(
+                    0, self.sentinel_token_ids.shape[-1], 1):
+                if torch.all(torch.eq(self.sentinel_token_ids, window)):
+                    return True
+        return False
+
+class Stream(transformers.StoppingCriteria):
+    def __init__(self, callback_func=None):
+        self.callback_func = callback_func
+
+    def __call__(self, input_ids, scores) -> bool:
+        if self.callback_func is not None:
+            self.callback_func(input_ids[0])
+        return False
+
+class Iteratorize:
+
+    """
+    Transforms a function that takes a callback
+    into a lazy iterator (generator).
+    """
+
+    def __init__(self, func, kwargs={}, callback=None):
+        self.mfunc=func
+        self.c_callback=callback
+        self.q = Queue()
+        self.sentinel = object()
+        self.kwargs = kwargs
+        self.stop_now = False
+
+        def _callback(val):
+            if self.stop_now:
+                raise ValueError
+            self.q.put(val)
+
+        def gentask():
+            try:
+                ret = self.mfunc(callback=_callback, **self.kwargs)
+            except ValueError:
+                pass
+            clear_torch_cache()
+            self.q.put(self.sentinel)
+            if self.c_callback:
+                self.c_callback(ret)
+
+        self.thread = Thread(target=gentask)
+        self.thread.start()
+
+    def __iter__(self):
+        return self
+
+    def __next__(self):
+        obj = self.q.get(True,None)
+        if obj is self.sentinel:
+            raise StopIteration
+        else:
+            return obj
+
+    def __del__(self):
+        clear_torch_cache()
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        self.stop_now = True
+        clear_torch_cache()
+
+def clear_torch_cache():
+    gc.collect()
+    if not shared.args.cpu:
+        torch.cuda.empty_cache()

modules/chat.py

@@ -84,6 +84,7 @@ def extract_message_from_reply(question, reply, name1, name2, check, impersonate
         tmp = f"\n{asker}:"
         for j in range(1, len(tmp)):
             if reply[-j:] == tmp[:j]:
+                reply = reply[:-j]
                 substring_found = True
 
     return reply, next_character_found, substring_found
@@ -91,7 +92,7 @@ def extract_message_from_reply(question, reply, name1, name2, check, impersonate
 def stop_everything_event():
     shared.stop_everything = True
 
-def chatbot_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typical_p, repetition_penalty, top_k, min_length, no_repeat_ngram_size, num_beams, penalty_alpha, length_penalty, early_stopping, name1, name2, context, check, chat_prompt_size, chat_generation_attempts=1):
+def chatbot_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typical_p, repetition_penalty, top_k, min_length, no_repeat_ngram_size, num_beams, penalty_alpha, length_penalty, early_stopping, name1, name2, context, check, chat_prompt_size, chat_generation_attempts=1, regenerate=False):
     shared.stop_everything = False
     just_started = True
     eos_token = '\n' if check else None
@@ -120,6 +121,10 @@ def chatbot_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typical
     else:
         prompt = custom_generate_chat_prompt(text, max_new_tokens, name1, name2, context, chat_prompt_size)
 
+    if not regenerate:
+        # Display user input and "*is typing...*" imediately
+        yield shared.history['visible']+[[visible_text, '*Is typing...*']]
+
     # Generate
     reply = ''
     for i in range(chat_generation_attempts):
@@ -158,6 +163,9 @@ def impersonate_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typ
 
     prompt = generate_chat_prompt(text, max_new_tokens, name1, name2, context, chat_prompt_size, impersonate=True)
 
+    # Display "*is typing...*" imediately
+    yield '*Is typing...*'
+
     reply = ''
     for i in range(chat_generation_attempts):
         for reply in generate_reply(prompt+reply, max_new_tokens, do_sample, temperature, top_p, typical_p, repetition_penalty, top_k, min_length, no_repeat_ngram_size, num_beams, penalty_alpha, length_penalty, early_stopping, eos_token=eos_token, stopping_string=f"\n{name2}:"):
@@ -182,7 +190,7 @@ def regenerate_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typi
         last_visible = shared.history['visible'].pop()
         last_internal = shared.history['internal'].pop()
 
-        for _history in chatbot_wrapper(last_internal[0], max_new_tokens, do_sample, temperature, top_p, typical_p, repetition_penalty, top_k, min_length, no_repeat_ngram_size, num_beams, penalty_alpha, length_penalty, early_stopping, name1, name2, context, check, chat_prompt_size, chat_generation_attempts):
+        for _history in chatbot_wrapper(last_internal[0], max_new_tokens, do_sample, temperature, top_p, typical_p, repetition_penalty, top_k, min_length, no_repeat_ngram_size, num_beams, penalty_alpha, length_penalty, early_stopping, name1, name2, context, check, chat_prompt_size, chat_generation_attempts, regenerate=True):
             if shared.args.cai_chat:
                 shared.history['visible'][-1] = [last_visible[0], _history[-1][1]]
                 yield generate_chat_html(shared.history['visible'], name1, name2, shared.character)
@@ -291,7 +299,7 @@ def save_history(timestamp=True):
         fname = f"{prefix}persistent.json"
     if not Path('logs').exists():
         Path('logs').mkdir()
-    with open(Path(f'logs/{fname}'), 'w') as f:
+    with open(Path(f'logs/{fname}'), 'w', encoding='utf-8') as f:
         f.write(json.dumps({'data': shared.history['internal'], 'data_visible': shared.history['visible']}, indent=2))
     return Path(f'logs/{fname}')
 
@@ -332,7 +340,7 @@ def load_character(_character, name1, name2):
     shared.history['visible'] = []
     if _character != 'None':
         shared.character = _character
-        data = json.loads(open(Path(f'characters/{_character}.json'), 'r').read())
+        data = json.loads(open(Path(f'characters/{_character}.json'), 'r', encoding='utf-8').read())
         name2 = data['char_name']
         if 'char_persona' in data and data['char_persona'] != '':
             context += f"{data['char_name']}'s Persona: {data['char_persona']}\n"
@@ -372,7 +380,7 @@ def upload_character(json_file, img, tavern=False):
         i += 1
     if tavern:
         outfile_name = f'TavernAI-{outfile_name}'
-    with open(Path(f'characters/{outfile_name}.json'), 'w') as f:
+    with open(Path(f'characters/{outfile_name}.json'), 'w', encoding='utf-8') as f:
         f.write(json_file)
     if img is not None:
         img = Image.open(io.BytesIO(img))

modules/models.py

@@ -1,5 +1,6 @@
 import json
 import os
+import sys
 import time
 import zipfile
 from pathlib import Path
@@ -41,7 +42,7 @@ def load_model(model_name):
     shared.is_RWKV = model_name.lower().startswith('rwkv-')
 
     # Default settings
-    if not (shared.args.cpu or shared.args.load_in_8bit or shared.args.auto_devices or shared.args.disk or shared.args.gpu_memory is not None or shared.args.cpu_memory is not None or shared.args.deepspeed or shared.args.flexgen or shared.is_RWKV):
+    if not any([shared.args.cpu, shared.args.load_in_8bit, shared.args.load_in_4bit, shared.args.gptq_bits > 0, shared.args.auto_devices, shared.args.disk, shared.args.gpu_memory is not None, shared.args.cpu_memory is not None, shared.args.deepspeed, shared.args.flexgen, shared.is_RWKV]):
         if any(size in shared.model_name.lower() for size in ('13b', '20b', '30b')):
             model = AutoModelForCausalLM.from_pretrained(Path(f"models/{shared.model_name}"), device_map='auto', load_in_8bit=True)
         else:
@@ -86,6 +87,12 @@ def load_model(model_name):
 
         return model, tokenizer
 
+    # 4-bit LLaMA
+    elif shared.args.gptq_bits > 0 or shared.args.load_in_4bit:
+        from modules.quantized_LLaMA import load_quantized_LLaMA
+
+        model = load_quantized_LLaMA(model_name)
+
     # Custom
     else:
         command = "AutoModelForCausalLM.from_pretrained"

modules/quantized_LLaMA.py

@@ -0,0 +1,60 @@
+import os
+import sys
+from pathlib import Path
+
+import accelerate
+import torch
+
+import modules.shared as shared
+
+sys.path.insert(0, os.path.abspath(Path("repositories/GPTQ-for-LLaMa")))
+from llama import load_quant
+
+
+# 4-bit LLaMA
+def load_quantized_LLaMA(model_name):
+    if shared.args.load_in_4bit:
+        bits = 4
+    else:
+        bits = shared.args.gptq_bits
+
+    path_to_model = Path(f'models/{model_name}')
+    pt_model = ''
+    if path_to_model.name.lower().startswith('llama-7b'):
+        pt_model = f'llama-7b-{bits}bit.pt'
+    elif path_to_model.name.lower().startswith('llama-13b'):
+        pt_model = f'llama-13b-{bits}bit.pt'
+    elif path_to_model.name.lower().startswith('llama-30b'):
+        pt_model = f'llama-30b-{bits}bit.pt'
+    elif path_to_model.name.lower().startswith('llama-65b'):
+        pt_model = f'llama-65b-{bits}bit.pt'
+    else:
+        pt_model = f'{model_name}-{bits}bit.pt'
+
+    # Try to find the .pt both in models/ and in the subfolder
+    pt_path = None
+    for path in [Path(p) for p in [f"models/{pt_model}", f"{path_to_model}/{pt_model}"]]:
+        if path.exists():
+            pt_path = path
+
+    if not pt_path:
+        print(f"Could not find {pt_model}, exiting...")
+        exit()
+
+    model = load_quant(path_to_model, os.path.abspath(pt_path), bits)
+
+    # Multi-GPU setup
+    if shared.args.gpu_memory:
+        max_memory = {}
+        for i in range(len(shared.args.gpu_memory)):
+            max_memory[i] = f"{shared.args.gpu_memory[i]}GiB"
+        max_memory['cpu'] = f"{shared.args.cpu_memory or '99'}GiB"
+
+        device_map = accelerate.infer_auto_device_map(model, max_memory=max_memory, no_split_module_classes=["LLaMADecoderLayer"])
+        model = accelerate.dispatch_model(model, device_map=device_map)
+
+    # Single GPU
+    else:
+        model = model.to(torch.device('cuda:0'))
+
+    return model

modules/shared.py

@@ -11,6 +11,7 @@ is_RWKV = False
 history = {'internal': [], 'visible': []}
 character = 'None'
 stop_everything = False
+still_streaming = False
 
 # UI elements (buttons, sliders, HTML, etc)
 gradio = {}
@@ -42,12 +43,12 @@ settings = {
         'default': 'NovelAI-Sphinx Moth',
         'pygmalion-*': 'Pygmalion',
         'RWKV-*': 'Naive',
-        '(rosey|chip|joi)_.*_instruct.*': 'Instruct Joi (Contrastive Search)'
     },
     'prompts': {
         'default': 'Common sense questions and answers\n\nQuestion: \nFactual answer:',
         '^(gpt4chan|gpt-4chan|4chan)': '-----\n--- 865467536\nInput text\n--- 865467537\n',
-        '(rosey|chip|joi)_.*_instruct.*': 'User: \n'
+        '(rosey|chip|joi)_.*_instruct.*': 'User: \n',
+        'oasst-*': '<|prompter|>Write a story about future of AI development<|endoftext|><|assistant|>'
     }
 }
 
@@ -68,6 +69,8 @@ parser.add_argument('--chat', action='store_true', help='Launch the web UI in ch
 parser.add_argument('--cai-chat', action='store_true', help='Launch the web UI in chat mode with a style similar to Character.AI\'s. If the file img_bot.png or img_bot.jpg exists in the same folder as server.py, this image will be used as the bot\'s profile picture. Similarly, img_me.png or img_me.jpg will be used as your profile picture.')
 parser.add_argument('--cpu', action='store_true', help='Use the CPU to generate text.')
 parser.add_argument('--load-in-8bit', action='store_true', help='Load the model with 8-bit precision.')
+parser.add_argument('--load-in-4bit', action='store_true', help='Load the model with 4-bit precision. Currently only works with LLaMA.')
+parser.add_argument('--gptq-bits', type=int, default=0, help='Load a pre-quantized model with specified precision. 2, 3, 4 and 8bit are supported. Currently only works with LLaMA.')
 parser.add_argument('--bf16', action='store_true', help='Load the model with bfloat16 precision. Requires NVIDIA Ampere GPU.')
 parser.add_argument('--auto-devices', action='store_true', help='Automatically split the model across the available GPU(s) and CPU.')
 parser.add_argument('--disk', action='store_true', help='If the model is too large for your GPU(s) and CPU combined, send the remaining layers to the disk.')
@@ -90,4 +93,5 @@ parser.add_argument('--listen', action='store_true', help='Make the web UI reach
 parser.add_argument('--listen-port', type=int, help='The listening port that the server will use.')
 parser.add_argument('--share', action='store_true', help='Create a public URL. This is useful for running the web UI on Google Colab or similar.')
 parser.add_argument('--verbose', action='store_true', help='Print the prompts to the terminal.')
+parser.add_argument('--auto-launch', action='store_true', default=False, help='Open the web UI in the default browser upon launch')
 args = parser.parse_args()

modules/stopping_criteria.py

@@ -1,32 +0,0 @@
-'''
-This code was copied from
-
-https://github.com/PygmalionAI/gradio-ui/
-
-'''
-
-import torch
-import transformers
-
-
-class _SentinelTokenStoppingCriteria(transformers.StoppingCriteria):
-
-    def __init__(self, sentinel_token_ids: torch.LongTensor,
-                 starting_idx: int):
-        transformers.StoppingCriteria.__init__(self)
-        self.sentinel_token_ids = sentinel_token_ids
-        self.starting_idx = starting_idx
-
-    def __call__(self, input_ids: torch.LongTensor,
-                 _scores: torch.FloatTensor) -> bool:
-        for sample in input_ids:
-            trimmed_sample = sample[self.starting_idx:]
-            # Can't unfold, output is still too tiny. Skip.
-            if trimmed_sample.shape[-1] < self.sentinel_token_ids.shape[-1]:
-                continue
-
-            for window in trimmed_sample.unfold(
-                    0, self.sentinel_token_ids.shape[-1], 1):
-                if torch.all(torch.eq(self.sentinel_token_ids, window)):
-                    return True
-        return False

modules/text_generation.py

@@ -5,13 +5,13 @@ import time
 import numpy as np
 import torch
 import transformers
-from tqdm import tqdm
 
 import modules.shared as shared
+from modules.callbacks import (Iteratorize, Stream,
+                               _SentinelTokenStoppingCriteria)
 from modules.extensions import apply_extensions
 from modules.html_generator import generate_4chan_html, generate_basic_html
 from modules.models import local_rank
-from modules.stopping_criteria import _SentinelTokenStoppingCriteria
 
 
 def get_max_prompt_length(tokens):
@@ -92,19 +92,22 @@ def generate_reply(question, max_new_tokens, do_sample, temperature, top_p, typi
     # These models are not part of Hugging Face, so we handle them
     # separately and terminate the function call earlier
     if shared.is_RWKV:
-        if shared.args.no_stream:
-            reply = shared.model.generate(context=question, token_count=max_new_tokens, temperature=temperature, top_p=top_p, top_k=top_k)
-            yield formatted_outputs(reply, shared.model_name)
-        else:
-            yield formatted_outputs(question, shared.model_name)
-            # RWKV has proper streaming, which is very nice.
-            # No need to generate 8 tokens at a time.
-            for reply in shared.model.generate_with_streaming(context=question, token_count=max_new_tokens, temperature=temperature, top_p=top_p, top_k=top_k):
+        try:
+            if shared.args.no_stream:
+                reply = shared.model.generate(context=question, token_count=max_new_tokens, temperature=temperature, top_p=top_p, top_k=top_k)
                 yield formatted_outputs(reply, shared.model_name)
-
-        t1 = time.time()
-        print(f"Output generated in {(t1-t0):.2f} seconds.")
-        return
+            else:
+                yield formatted_outputs(question, shared.model_name)
+                # RWKV has proper streaming, which is very nice.
+                # No need to generate 8 tokens at a time.
+                for reply in shared.model.generate_with_streaming(context=question, token_count=max_new_tokens, temperature=temperature, top_p=top_p, top_k=top_k):
+                    yield formatted_outputs(reply, shared.model_name)
+        finally:
+            t1 = time.time()
+            output = encode(reply)[0]
+            input_ids = encode(question)
+            print(f"Output generated in {(t1-t0):.2f} seconds ({(len(output)-len(input_ids[0]))/(t1-t0):.2f} tokens/s, {len(output)-len(input_ids[0])} tokens)")
+            return
 
     original_question = question
     if not (shared.args.chat or shared.args.cai_chat):
@@ -113,24 +116,22 @@ def generate_reply(question, max_new_tokens, do_sample, temperature, top_p, typi
         print(f"\n\n{question}\n--------------------\n")
 
     input_ids = encode(question, max_new_tokens)
+    original_input_ids = input_ids
+    output = input_ids[0]
     cuda = "" if (shared.args.cpu or shared.args.deepspeed or shared.args.flexgen) else ".cuda()"
-    n = shared.tokenizer.eos_token_id if eos_token is None else int(encode(eos_token)[0][-1])
+    eos_token_ids = [shared.tokenizer.eos_token_id]
+    if eos_token is not None:
+        eos_token_ids.append(int(encode(eos_token)[0][-1]))
+    stopping_criteria_list = transformers.StoppingCriteriaList()
     if stopping_string is not None:
-        # The stopping_criteria code below was copied from
-        # https://github.com/PygmalionAI/gradio-ui/blob/master/src/model.py
+        # Copied from https://github.com/PygmalionAI/gradio-ui/blob/master/src/model.py
         t = encode(stopping_string, 0, add_special_tokens=False)
-        stopping_criteria_list = transformers.StoppingCriteriaList([
-            _SentinelTokenStoppingCriteria(
-                sentinel_token_ids=t,
-                starting_idx=len(input_ids[0])
-            )
-        ])
-    else:
-        stopping_criteria_list = None
+        stopping_criteria_list.append(_SentinelTokenStoppingCriteria(sentinel_token_ids=t, starting_idx=len(input_ids[0])))
 
     if not shared.args.flexgen:
         generate_params = [
-            f"eos_token_id={n}",
+            f"max_new_tokens=max_new_tokens",
+            f"eos_token_id={eos_token_ids}",
             f"stopping_criteria=stopping_criteria_list",
             f"do_sample={do_sample}",
             f"temperature={temperature}",
@@ -147,44 +148,23 @@ def generate_reply(question, max_new_tokens, do_sample, temperature, top_p, typi
         ]
     else:
         generate_params = [
+            f"max_new_tokens={max_new_tokens if shared.args.no_stream else 8}",
             f"do_sample={do_sample}",
             f"temperature={temperature}",
-            f"stop={n}",
+            f"stop={eos_token_ids[-1]}",
         ]
     if shared.args.deepspeed:
         generate_params.append("synced_gpus=True")
-    if shared.args.no_stream:
-        generate_params.append("max_new_tokens=max_new_tokens")
-    else:
-        generate_params.append("max_new_tokens=8")
     if shared.soft_prompt:
         inputs_embeds, filler_input_ids = generate_softprompt_input_tensors(input_ids)
         generate_params.insert(0, "inputs_embeds=inputs_embeds")
-        generate_params.insert(0, "filler_input_ids")
+        generate_params.insert(0, "inputs=filler_input_ids")
     else:
-        generate_params.insert(0, "input_ids")
-
-    # Generate the entire reply at once
-    if shared.args.no_stream:
-        with torch.no_grad():
-            output = eval(f"shared.model.generate({', '.join(generate_params)}){cuda}")[0]
-        if shared.soft_prompt:
-            output = torch.cat((input_ids[0], output[filler_input_ids.shape[1]:]))
-
-        reply = decode(output)
-        if not (shared.args.chat or shared.args.cai_chat):
-            reply = original_question + apply_extensions(reply[len(question):], "output")
-
-        t1 = time.time()
-        print(f"Output generated in {(t1-t0):.2f} seconds ({(len(output)-len(input_ids[0]))/(t1-t0)/8:.2f} it/s, {len(output)-len(input_ids[0])} tokens)")
-        yield formatted_outputs(reply, shared.model_name)
-
-    # Generate the reply 8 tokens at a time
-    else:
-        yield formatted_outputs(original_question, shared.model_name)
-        for i in tqdm(range(max_new_tokens//8+1)):
-            clear_torch_cache()
+        generate_params.insert(0, "inputs=input_ids")
 
+    try:
+        # Generate the entire reply at once.
+        if shared.args.no_stream:
             with torch.no_grad():
                 output = eval(f"shared.model.generate({', '.join(generate_params)}){cuda}")[0]
             if shared.soft_prompt:
@@ -193,16 +173,66 @@ def generate_reply(question, max_new_tokens, do_sample, temperature, top_p, typi
             reply = decode(output)
             if not (shared.args.chat or shared.args.cai_chat):
                 reply = original_question + apply_extensions(reply[len(question):], "output")
+
             yield formatted_outputs(reply, shared.model_name)
 
-            if not shared.args.flexgen:
-                if output[-1] == n:
-                    break
-                input_ids = torch.reshape(output, (1, output.shape[0]))
-            else:
-                if np.count_nonzero(input_ids[0] == n) < np.count_nonzero(output == n):
+        # Stream the reply 1 token at a time.
+        # This is based on the trick of using 'stopping_criteria' to create an iterator.
+        elif not shared.args.flexgen:
+
+            def generate_with_callback(callback=None, **kwargs):
+                kwargs['stopping_criteria'].append(Stream(callback_func=callback))
+                clear_torch_cache()
+                with torch.no_grad():
+                    shared.model.generate(**kwargs)
+
+            def generate_with_streaming(**kwargs):
+                return Iteratorize(generate_with_callback, kwargs, callback=None)
+
+            shared.still_streaming = True
+            yield formatted_outputs(original_question, shared.model_name)
+            with eval(f"generate_with_streaming({', '.join(generate_params)})") as generator:
+                for output in generator:
+                    if shared.soft_prompt:
+                        output = torch.cat((input_ids[0], output[filler_input_ids.shape[1]:]))
+                    reply = decode(output)
+
+                    if not (shared.args.chat or shared.args.cai_chat):
+                        reply = original_question + apply_extensions(reply[len(question):], "output")
+
+                    if output[-1] in eos_token_ids:
+                        break
+                    yield formatted_outputs(reply, shared.model_name)
+
+                shared.still_streaming = False
+                yield formatted_outputs(reply, shared.model_name)
+
+        # Stream the output naively for FlexGen since it doesn't support 'stopping_criteria'
+        else:
+            shared.still_streaming = True
+            for i in range(max_new_tokens//8+1):
+                clear_torch_cache()
+                with torch.no_grad():
+                    output = eval(f"shared.model.generate({', '.join(generate_params)})")[0]
+                if shared.soft_prompt:
+                    output = torch.cat((input_ids[0], output[filler_input_ids.shape[1]:]))
+                reply = decode(output)
+
+                if not (shared.args.chat or shared.args.cai_chat):
+                    reply = original_question + apply_extensions(reply[len(question):], "output")
+
+                if np.count_nonzero(np.isin(input_ids[0], eos_token_ids)) < np.count_nonzero(np.isin(output, eos_token_ids)):
                     break
+                yield formatted_outputs(reply, shared.model_name)
+
                 input_ids = np.reshape(output, (1, output.shape[0]))
+                if shared.soft_prompt:
+                    inputs_embeds, filler_input_ids = generate_softprompt_input_tensors(input_ids)
 
-            if shared.soft_prompt:
-                inputs_embeds, filler_input_ids = generate_softprompt_input_tensors(input_ids)
+            shared.still_streaming = False
+            yield formatted_outputs(reply, shared.model_name)
+
+    finally:
+        t1 = time.time()
+        print(f"Output generated in {(t1-t0):.2f} seconds ({(len(output)-len(original_input_ids[0]))/(t1-t0):.2f} tokens/s, {len(output)-len(original_input_ids[0])} tokens)")
+        return

requirements.txt

@@ -1,9 +1,11 @@
-accelerate==0.16.0
+accelerate==0.17.0
 bitsandbytes==0.37.0
 flexgen==0.1.7
 gradio==3.18.0
 numpy
-rwkv==0.1.0
-safetensors==0.2.8
+requests
+rwkv==0.3.1
+safetensors==0.3.0
 sentencepiece
-git+https://github.com/oobabooga/transformers@llama_push
+tqdm
+git+https://github.com/zphang/transformers@llama_push

server.py

@@ -18,9 +18,6 @@ from modules.html_generator import generate_chat_html
 from modules.models import load_model, load_soft_prompt
 from modules.text_generation import generate_reply
 
-if (shared.args.chat or shared.args.cai_chat) and not shared.args.no_stream:
-    print('Warning: chat mode currently becomes somewhat slower with text streaming on.\nConsider starting the web UI with the --no-stream option.\n')
-    
 # Loading custom settings
 settings_file = None
 if shared.args.settings is not None and Path(shared.args.settings).exists():
@@ -37,7 +34,7 @@ def get_available_models():
     if shared.args.flexgen:
         return sorted([re.sub('-np$', '', item.name) for item in list(Path('models/').glob('*')) if item.name.endswith('-np')], key=str.lower)
     else:
-        return sorted([item.name for item in list(Path('models/').glob('*')) if not item.name.endswith(('.txt', '-np'))], key=str.lower)
+        return sorted([item.name for item in list(Path('models/').glob('*')) if not item.name.endswith(('.txt', '-np', '.pt'))], key=str.lower)
 
 def get_available_presets():
     return sorted(set(map(lambda x : '.'.join(str(x.name).split('.')[:-1]), Path('presets').glob('*.txt'))), key=str.lower)
@@ -272,10 +269,10 @@ if shared.args.chat or shared.args.cai_chat:
 
         function_call = 'chat.cai_chatbot_wrapper' if shared.args.cai_chat else 'chat.chatbot_wrapper'
 
-        gen_events.append(shared.gradio['Generate'].click(eval(function_call), shared.input_params, shared.gradio['display'], show_progress=shared.args.no_stream, api_name='textgen'))
-        gen_events.append(shared.gradio['textbox'].submit(eval(function_call), shared.input_params, shared.gradio['display'], show_progress=shared.args.no_stream))
-        gen_events.append(shared.gradio['Regenerate'].click(chat.regenerate_wrapper, shared.input_params, shared.gradio['display'], show_progress=shared.args.no_stream))
-        gen_events.append(shared.gradio['Impersonate'].click(chat.impersonate_wrapper, shared.input_params, shared.gradio['textbox'], show_progress=shared.args.no_stream))
+        gen_events.append(shared.gradio['Generate'].click(eval(function_call), shared.input_params, shared.gradio['display'], show_progress=False, api_name='textgen'))
+        gen_events.append(shared.gradio['textbox'].submit(eval(function_call), shared.input_params, shared.gradio['display'], show_progress=False))
+        gen_events.append(shared.gradio['Regenerate'].click(chat.regenerate_wrapper, shared.input_params, shared.gradio['display'], show_progress=False))
+        gen_events.append(shared.gradio['Impersonate'].click(chat.impersonate_wrapper, shared.input_params, shared.gradio['textbox'], show_progress=False))
         shared.gradio['Stop'].click(chat.stop_everything_event, [], [], cancels=gen_events)
 
         shared.gradio['Copy last reply'].click(chat.send_last_reply_to_input, [], shared.gradio['textbox'], show_progress=shared.args.no_stream)
@@ -309,6 +306,7 @@ if shared.args.chat or shared.args.cai_chat:
         reload_inputs = [shared.gradio['name1'], shared.gradio['name2']] if shared.args.cai_chat else []
         shared.gradio['upload_chat_history'].upload(reload_func, reload_inputs, [shared.gradio['display']])
         shared.gradio['upload_img_me'].upload(reload_func, reload_inputs, [shared.gradio['display']])
+        shared.gradio['Stop'].click(reload_func, reload_inputs, [shared.gradio['display']])
 
         shared.gradio['interface'].load(lambda : chat.load_default_history(shared.settings[f'name1{suffix}'], shared.settings[f'name2{suffix}']), None, None)
         shared.gradio['interface'].load(reload_func, reload_inputs, [shared.gradio['display']], show_progress=True)
@@ -372,9 +370,9 @@ else:
 
 shared.gradio['interface'].queue()
 if shared.args.listen:
-    shared.gradio['interface'].launch(prevent_thread_lock=True, share=shared.args.share, server_name='0.0.0.0', server_port=shared.args.listen_port)
+    shared.gradio['interface'].launch(prevent_thread_lock=True, share=shared.args.share, server_name='0.0.0.0', server_port=shared.args.listen_port, inbrowser=shared.args.auto_launch)
 else:
-    shared.gradio['interface'].launch(prevent_thread_lock=True, share=shared.args.share, server_port=shared.args.listen_port)
+    shared.gradio['interface'].launch(prevent_thread_lock=True, share=shared.args.share, server_port=shared.args.listen_port, inbrowser=shared.args.auto_launch)
 
 # I think that I will need this later
 while True:

settings-template.json

@@ -29,6 +29,7 @@
     "prompts": {
         "default": "Common sense questions and answers\n\nQuestion: \nFactual answer:",
         "^(gpt4chan|gpt-4chan|4chan)": "-----\n--- 865467536\nInput text\n--- 865467537\n",
-        "(rosey|chip|joi)_.*_instruct.*": "User: \n"
+        "(rosey|chip|joi)_.*_instruct.*": "User: \n",
+        "oasst-*": "<|prompter|>Write a story about future of AI development<|endoftext|><|assistant|>"
     }
 }