diff --git a/download-model.py b/download-model.py
index 82e956d6..d7cf9273 100644
--- a/download-model.py
+++ b/download-model.py
@@ -26,13 +26,16 @@ base = "https://huggingface.co"
 
 class ModelDownloader:
     def __init__(self, max_retries=5):
-        self.session = requests.Session()
-        if max_retries:
-            self.session.mount('https://cdn-lfs.huggingface.co', HTTPAdapter(max_retries=max_retries))
-            self.session.mount('https://huggingface.co', HTTPAdapter(max_retries=max_retries))
+        self.max_retries = max_retries
+
+    def get_session(self):
+        session = requests.Session()
+        if self.max_retries:
+            session.mount('https://cdn-lfs.huggingface.co', HTTPAdapter(max_retries=self.max_retries))
+            session.mount('https://huggingface.co', HTTPAdapter(max_retries=self.max_retries))
 
         if os.getenv('HF_USER') is not None and os.getenv('HF_PASS') is not None:
-            self.session.auth = (os.getenv('HF_USER'), os.getenv('HF_PASS'))
+            session.auth = (os.getenv('HF_USER'), os.getenv('HF_PASS'))
 
         try:
             from huggingface_hub import get_token
@@ -41,7 +44,9 @@ class ModelDownloader:
             token = os.getenv("HF_TOKEN")
 
         if token is not None:
-            self.session.headers = {'authorization': f'Bearer {token}'}
+            session.headers = {'authorization': f'Bearer {token}'}
+
+        return session
 
     def sanitize_model_and_branch_names(self, model, branch):
         if model[-1] == '/':
@@ -65,6 +70,7 @@ class ModelDownloader:
         return model, branch
 
     def get_download_links_from_huggingface(self, model, branch, text_only=False, specific_file=None):
+        session = self.get_session()
         page = f"/api/models/{model}/tree/{branch}"
         cursor = b""
 
@@ -78,7 +84,7 @@ class ModelDownloader:
         is_lora = False
         while True:
             url = f"{base}{page}" + (f"?cursor={cursor.decode()}" if cursor else "")
-            r = self.session.get(url, timeout=10)
+            r = session.get(url, timeout=10)
             r.raise_for_status()
             content = r.content
 
@@ -156,9 +162,8 @@ class ModelDownloader:
         is_llamacpp = has_gguf and specific_file is not None
         return links, sha256, is_lora, is_llamacpp
 
-    def get_output_folder(self, model, branch, is_lora, is_llamacpp=False, base_folder=None):
-        if base_folder is None:
-            base_folder = 'models' if not is_lora else 'loras'
+    def get_output_folder(self, model, branch, is_lora, is_llamacpp=False):
+        base_folder = 'models' if not is_lora else 'loras'
 
         # If the model is of type GGUF, save directly in the base_folder
         if is_llamacpp:
@@ -172,6 +177,7 @@ class ModelDownloader:
         return output_folder
 
     def get_single_file(self, url, output_folder, start_from_scratch=False):
+        session = self.get_session()
         filename = Path(url.rsplit('/', 1)[1])
         output_path = output_folder / filename
         headers = {}
@@ -179,7 +185,7 @@ class ModelDownloader:
         if output_path.exists() and not start_from_scratch:
 
             # Check if the file has already been downloaded completely
-            r = self.session.get(url, stream=True, timeout=10)
+            r = session.get(url, stream=True, timeout=10)
             total_size = int(r.headers.get('content-length', 0))
             if output_path.stat().st_size >= total_size:
                 return
@@ -188,7 +194,7 @@ class ModelDownloader:
             headers = {'Range': f'bytes={output_path.stat().st_size}-'}
             mode = 'ab'
 
-        with self.session.get(url, stream=True, headers=headers, timeout=10) as r:
+        with session.get(url, stream=True, headers=headers, timeout=10) as r:
             r.raise_for_status()  # Do not continue the download if the request was unsuccessful
             total_size = int(r.headers.get('content-length', 0))
             block_size = 1024 * 1024  # 1MB
@@ -303,7 +309,10 @@ if __name__ == '__main__':
     links, sha256, is_lora, is_llamacpp = downloader.get_download_links_from_huggingface(model, branch, text_only=args.text_only, specific_file=specific_file)
 
     # Get the output folder
-    output_folder = downloader.get_output_folder(model, branch, is_lora, is_llamacpp=is_llamacpp, base_folder=args.output)
+    if args.output:
+        output_folder = Path(args.output)
+    else:
+        output_folder = downloader.get_output_folder(model, branch, is_lora, is_llamacpp=is_llamacpp)
 
     if args.check:
         # Check previously downloaded files
diff --git a/instruction-templates/Mistral.yaml b/instruction-templates/Mistral.yaml
index 113c238b..02e5b20d 100644
--- a/instruction-templates/Mistral.yaml
+++ b/instruction-templates/Mistral.yaml
@@ -4,7 +4,7 @@ instruction_template: |-
           {{- message['content'] -}}
       {%- else -%}
           {%- if message['role'] == 'user' -%}
-              {{-' [INST] ' + message['content'].rstrip() + ' [/INST] '-}}
+              {{-'[INST] ' + message['content'].rstrip() + ' [/INST]'-}}
           {%- else -%}
               {{-'' + message['content'] + '</s>' -}}
           {%- endif -%}
diff --git a/modules/chat.py b/modules/chat.py
index bddc3132..de7f19de 100644
--- a/modules/chat.py
+++ b/modules/chat.py
@@ -166,53 +166,54 @@ def generate_chat_prompt(user_input, state, **kwargs):
         prompt = remove_extra_bos(prompt)
         return prompt
 
-    # Handle truncation
-    max_length = get_max_prompt_length(state)
     prompt = make_prompt(messages)
-    encoded_length = get_encoded_length(prompt)
 
-    while len(messages) > 0 and encoded_length > max_length:
+    # Handle truncation
+    if shared.tokenizer is not None:
+        max_length = get_max_prompt_length(state)
+        encoded_length = get_encoded_length(prompt)
+        while len(messages) > 0 and encoded_length > max_length:
 
-        # Remove old message, save system message
-        if len(messages) > 2 and messages[0]['role'] == 'system':
-            messages.pop(1)
+            # Remove old message, save system message
+            if len(messages) > 2 and messages[0]['role'] == 'system':
+                messages.pop(1)
 
-        # Remove old message when no system message is present
-        elif len(messages) > 1 and messages[0]['role'] != 'system':
-            messages.pop(0)
+            # Remove old message when no system message is present
+            elif len(messages) > 1 and messages[0]['role'] != 'system':
+                messages.pop(0)
 
-        # Resort to truncating the user input
-        else:
+            # Resort to truncating the user input
+            else:
 
-            user_message = messages[-1]['content']
+                user_message = messages[-1]['content']
 
-            # Bisect the truncation point
-            left, right = 0, len(user_message) - 1
+                # Bisect the truncation point
+                left, right = 0, len(user_message) - 1
 
-            while right - left > 1:
-                mid = (left + right) // 2
+                while right - left > 1:
+                    mid = (left + right) // 2
 
-                messages[-1]['content'] = user_message[mid:]
+                    messages[-1]['content'] = user_message[mid:]
+                    prompt = make_prompt(messages)
+                    encoded_length = get_encoded_length(prompt)
+
+                    if encoded_length <= max_length:
+                        right = mid
+                    else:
+                        left = mid
+
+                messages[-1]['content'] = user_message[right:]
                 prompt = make_prompt(messages)
                 encoded_length = get_encoded_length(prompt)
-
-                if encoded_length <= max_length:
-                    right = mid
+                if encoded_length > max_length:
+                    logger.error(f"Failed to build the chat prompt. The input is too long for the available context length.\n\nTruncation length: {state['truncation_length']}\nmax_new_tokens: {state['max_new_tokens']} (is it too high?)\nAvailable context length: {max_length}\n")
+                    raise ValueError
                 else:
-                    left = mid
+                    logger.warning(f"The input has been truncated. Context length: {state['truncation_length']}, max_new_tokens: {state['max_new_tokens']}, available context length: {max_length}.")
+                    break
 
-            messages[-1]['content'] = user_message[right:]
             prompt = make_prompt(messages)
             encoded_length = get_encoded_length(prompt)
-            if encoded_length > max_length:
-                logger.error(f"Failed to build the chat prompt. The input is too long for the available context length.\n\nTruncation length: {state['truncation_length']}\nmax_new_tokens: {state['max_new_tokens']} (is it too high?)\nAvailable context length: {max_length}\n")
-                raise ValueError
-            else:
-                logger.warning(f"The input has been truncated. Context length: {state['truncation_length']}, max_new_tokens: {state['max_new_tokens']}, available context length: {max_length}.")
-                break
-
-        prompt = make_prompt(messages)
-        encoded_length = get_encoded_length(prompt)
 
     if also_return_rows:
         return prompt, [message['content'] for message in messages]
@@ -690,6 +691,9 @@ def load_character(character, name1, name2):
 
 
 def load_instruction_template(template):
+    if template == 'None':
+        return ''
+
     for filepath in [Path(f'instruction-templates/{template}.yaml'), Path('instruction-templates/Alpaca.yaml')]:
         if filepath.exists():
             break
diff --git a/modules/exllamav2.py b/modules/exllamav2.py
index 551ed498..239c2031 100644
--- a/modules/exllamav2.py
+++ b/modules/exllamav2.py
@@ -51,18 +51,21 @@ class Exllamav2Model:
 
         model = ExLlamaV2(config)
 
-        split = None
-        if shared.args.gpu_split:
-            split = [float(alloc) for alloc in shared.args.gpu_split.split(",")]
+        if shared.args.cache_8bit:
+            cache = ExLlamaV2Cache_8bit(model, lazy=True)
+        else:
+            cache = ExLlamaV2Cache(model, lazy=True)
 
-        model.load(split)
+        if shared.args.autosplit:
+            model.load_autosplit(cache)
+        else:
+            split = None
+            if shared.args.gpu_split:
+                split = [float(alloc) for alloc in shared.args.gpu_split.split(",")]
+
+            model.load(split)
 
         tokenizer = ExLlamaV2Tokenizer(config)
-        if shared.args.cache_8bit:
-            cache = ExLlamaV2Cache_8bit(model)
-        else:
-            cache = ExLlamaV2Cache(model)
-
         generator = ExLlamaV2StreamingGenerator(model, cache, tokenizer)
 
         result = self()
diff --git a/modules/exllamav2_hf.py b/modules/exllamav2_hf.py
index 944c39dd..e5b35a44 100644
--- a/modules/exllamav2_hf.py
+++ b/modules/exllamav2_hf.py
@@ -37,18 +37,22 @@ class Exllamav2HF(PreTrainedModel):
         super().__init__(PretrainedConfig())
         self.ex_config = config
         self.ex_model = ExLlamaV2(config)
-        split = None
-        if shared.args.gpu_split:
-            split = [float(alloc) for alloc in shared.args.gpu_split.split(",")]
-
-        self.ex_model.load(split)
-        self.generation_config = GenerationConfig()
         self.loras = None
+        self.generation_config = GenerationConfig()
 
         if shared.args.cache_8bit:
-            self.ex_cache = ExLlamaV2Cache_8bit(self.ex_model)
+            self.ex_cache = ExLlamaV2Cache_8bit(self.ex_model, lazy=True)
         else:
-            self.ex_cache = ExLlamaV2Cache(self.ex_model)
+            self.ex_cache = ExLlamaV2Cache(self.ex_model, lazy=True)
+
+        if shared.args.autosplit:
+            self.ex_model.load_autosplit(self.ex_cache)
+        else:
+            split = None
+            if shared.args.gpu_split:
+                split = [float(alloc) for alloc in shared.args.gpu_split.split(",")]
+
+            self.ex_model.load(split)
 
         self.past_seq = None
         if shared.args.cfg_cache:
diff --git a/modules/loaders.py b/modules/loaders.py
index 26b7c5e2..08a7f229 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -78,6 +78,7 @@ loaders_and_params = OrderedDict({
         'no_flash_attn',
         'num_experts_per_token',
         'cache_8bit',
+        'autosplit',
         'alpha_value',
         'compress_pos_emb',
         'trust_remote_code',
@@ -89,6 +90,7 @@ loaders_and_params = OrderedDict({
         'no_flash_attn',
         'num_experts_per_token',
         'cache_8bit',
+        'autosplit',
         'alpha_value',
         'compress_pos_emb',
         'exllamav2_info',
diff --git a/modules/models.py b/modules/models.py
index d8f1a9f8..60568063 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -257,7 +257,7 @@ def llamacpp_HF_loader(model_name):
     path = Path(f'{shared.args.model_dir}/{model_name}')
 
     # Check if a HF tokenizer is available for the model
-    if all((path / file).exists() for file in ['tokenizer.model', 'tokenizer_config.json']):
+    if all((path / file).exists() for file in ['tokenizer_config.json']):
         logger.info(f'Using tokenizer from: \"{path}\"')
     else:
         logger.error("Could not load the model because a tokenizer in Transformers format was not found.")
diff --git a/modules/models_settings.py b/modules/models_settings.py
index 9acc7efa..659bc35d 100644
--- a/modules/models_settings.py
+++ b/modules/models_settings.py
@@ -153,6 +153,8 @@ def infer_loader(model_name, model_settings):
         loader = 'ExLlamav2_HF'
     elif (path_to_model / 'quant_config.json').exists() or re.match(r'.*-awq', model_name.lower()):
         loader = 'AutoAWQ'
+    elif len(list(path_to_model.glob('*.gguf'))) > 0 and path_to_model.is_dir() and (path_to_model / 'tokenizer_config.json').exists():
+        loader = 'llamacpp_HF'
     elif len(list(path_to_model.glob('*.gguf'))) > 0:
         loader = 'llama.cpp'
     elif re.match(r'.*\.gguf', model_name.lower()):
@@ -225,7 +227,7 @@ def apply_model_settings_to_state(model, state):
         loader = model_settings.pop('loader')
 
         # If the user is using an alternative loader for the same model type, let them keep using it
-        if not (loader == 'ExLlamav2_HF' and state['loader'] in ['GPTQ-for-LLaMa', 'ExLlamav2', 'AutoGPTQ']) and not (loader == 'llama.cpp' and state['loader'] in ['llamacpp_HF', 'ctransformers']):
+        if not (loader == 'ExLlamav2_HF' and state['loader'] in ['GPTQ-for-LLaMa', 'ExLlamav2', 'AutoGPTQ']) and not (loader == 'llama.cpp' and state['loader'] in ['ctransformers']):
             state['loader'] = loader
 
     for k in model_settings:
@@ -243,27 +245,54 @@ def save_model_settings(model, state):
     Save the settings for this model to models/config-user.yaml
     '''
     if model == 'None':
-        yield ("Not saving the settings because no model is loaded.")
+        yield ("Not saving the settings because no model is selected in the menu.")
         return
 
-    with Path(f'{shared.args.model_dir}/config-user.yaml') as p:
-        if p.exists():
-            user_config = yaml.safe_load(open(p, 'r').read())
-        else:
-            user_config = {}
+    user_config = shared.load_user_config()
+    model_regex = model + '$'  # For exact matches
+    if model_regex not in user_config:
+        user_config[model_regex] = {}
 
-        model_regex = model + '$'  # For exact matches
-        if model_regex not in user_config:
-            user_config[model_regex] = {}
+    for k in ui.list_model_elements():
+        if k == 'loader' or k in loaders.loaders_and_params[state['loader']]:
+            user_config[model_regex][k] = state[k]
 
-        for k in ui.list_model_elements():
-            if k == 'loader' or k in loaders.loaders_and_params[state['loader']]:
-                user_config[model_regex][k] = state[k]
+    shared.user_config = user_config
 
-        shared.user_config = user_config
+    output = yaml.dump(user_config, sort_keys=False)
+    p = Path(f'{shared.args.model_dir}/config-user.yaml')
+    with open(p, 'w') as f:
+        f.write(output)
 
-        output = yaml.dump(user_config, sort_keys=False)
-        with open(p, 'w') as f:
-            f.write(output)
+    yield (f"Settings for `{model}` saved to `{p}`.")
 
-        yield (f"Settings for `{model}` saved to `{p}`.")
+
+def save_instruction_template(model, template):
+    '''
+    Similar to the function above, but it saves only the instruction template.
+    '''
+    if model == 'None':
+        yield ("Not saving the template because no model is selected in the menu.")
+        return
+
+    user_config = shared.load_user_config()
+    model_regex = model + '$'  # For exact matches
+    if model_regex not in user_config:
+        user_config[model_regex] = {}
+
+    if template == 'None':
+        user_config[model_regex].pop('instruction_template', None)
+    else:
+        user_config[model_regex]['instruction_template'] = template
+
+    shared.user_config = user_config
+
+    output = yaml.dump(user_config, sort_keys=False)
+    p = Path(f'{shared.args.model_dir}/config-user.yaml')
+    with open(p, 'w') as f:
+        f.write(output)
+
+    if template == 'None':
+        yield (f"Instruction template for `{model}` unset in `{p}`, as the value for template was `{template}`.")
+    else:
+        yield (f"Instruction template for `{model}` saved to `{p}` as `{template}`.")
diff --git a/modules/shared.py b/modules/shared.py
index 2861d690..7bef04bf 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -134,6 +134,7 @@ group.add_argument('--row_split', action='store_true', help='Split the model by
 # ExLlamaV2
 group = parser.add_argument_group('ExLlamaV2')
 group.add_argument('--gpu-split', type=str, help='Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.')
+group.add_argument('--autosplit', action='store_true', help='Autosplit the model tensors across the available GPUs. This causes --gpu-split to be ignored.')
 group.add_argument('--max_seq_len', type=int, default=2048, help='Maximum sequence length.')
 group.add_argument('--cfg-cache', action='store_true', help='ExLlamav2_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.')
 group.add_argument('--no_flash_attn', action='store_true', help='Force flash-attention to not be used.')
@@ -279,6 +280,23 @@ def is_chat():
     return True
 
 
+def load_user_config():
+    '''
+    Loads custom model-specific settings
+    '''
+    if Path(f'{args.model_dir}/config-user.yaml').exists():
+        file_content = open(f'{args.model_dir}/config-user.yaml', 'r').read().strip()
+
+        if file_content:
+            user_config = yaml.safe_load(file_content)
+        else:
+            user_config = {}
+    else:
+        user_config = {}
+
+    return user_config
+
+
 args.loader = fix_loader_name(args.loader)
 
 # Activate the multimodal extension
@@ -297,11 +315,7 @@ with Path(f'{args.model_dir}/config.yaml') as p:
         model_config = {}
 
 # Load custom model-specific settings
-with Path(f'{args.model_dir}/config-user.yaml') as p:
-    if p.exists():
-        user_config = yaml.safe_load(open(p, 'r').read())
-    else:
-        user_config = {}
+user_config = load_user_config()
 
 model_config = OrderedDict(model_config)
 user_config = OrderedDict(user_config)
diff --git a/modules/ui.py b/modules/ui.py
index 06498f69..bb5a3339 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -76,6 +76,7 @@ def list_model_elements():
         'no_flash_attn',
         'num_experts_per_token',
         'cache_8bit',
+        'autosplit',
         'threads',
         'threads_batch',
         'n_batch',
diff --git a/modules/ui_chat.py b/modules/ui_chat.py
index a1b1af97..7576628d 100644
--- a/modules/ui_chat.py
+++ b/modules/ui_chat.py
@@ -109,7 +109,7 @@ def create_chat_settings_ui():
         with gr.Row():
             with gr.Column():
                 with gr.Row():
-                    shared.gradio['instruction_template'] = gr.Dropdown(choices=utils.get_available_instruction_templates(), label='Saved instruction templates', value='Select template to load...', elem_classes='slim-dropdown')
+                    shared.gradio['instruction_template'] = gr.Dropdown(choices=utils.get_available_instruction_templates(), label='Saved instruction templates', info="After selecting the template, click on \"Load\" to load and apply it.", value='None', elem_classes='slim-dropdown')
                     ui.create_refresh_button(shared.gradio['instruction_template'], lambda: None, lambda: {'choices': utils.get_available_instruction_templates()}, 'refresh-button', interactive=not mu)
                     shared.gradio['load_template'] = gr.Button("Load", elem_classes='refresh-button')
                     shared.gradio['save_template'] = gr.Button('💾', elem_classes='refresh-button', interactive=not mu)
diff --git a/modules/ui_model_menu.py b/modules/ui_model_menu.py
index 387915b1..ac6a8a8f 100644
--- a/modules/ui_model_menu.py
+++ b/modules/ui_model_menu.py
@@ -17,6 +17,7 @@ from modules.models import load_model, unload_model
 from modules.models_settings import (
     apply_model_settings_to_state,
     get_model_metadata,
+    save_instruction_template,
     save_model_settings,
     update_model_parameters
 )
@@ -131,6 +132,7 @@ def create_ui():
                             shared.gradio['disk'] = gr.Checkbox(label="disk", value=shared.args.disk)
                             shared.gradio['bf16'] = gr.Checkbox(label="bf16", value=shared.args.bf16)
                             shared.gradio['cache_8bit'] = gr.Checkbox(label="cache_8bit", value=shared.args.cache_8bit, info='Use 8-bit cache to save VRAM.')
+                            shared.gradio['autosplit'] = gr.Checkbox(label="autosplit", value=shared.args.autosplit, info='Automatically split the model tensors across the available GPUs.')
                             shared.gradio['no_flash_attn'] = gr.Checkbox(label="no_flash_attn", value=shared.args.no_flash_attn, info='Force flash-attention to not be used.')
                             shared.gradio['cfg_cache'] = gr.Checkbox(label="cfg-cache", value=shared.args.cfg_cache, info='Necessary to use CFG with this loader.')
                             shared.gradio['num_experts_per_token'] = gr.Number(label="Number of experts per token", value=shared.args.num_experts_per_token, info='Only applies to MoE models like Mixtral.')
@@ -143,17 +145,35 @@ def create_ui():
                             shared.gradio['disable_exllamav2'] = gr.Checkbox(label="disable_exllamav2", value=shared.args.disable_exllamav2, info='Disable ExLlamav2 kernel for GPTQ models.')
                             shared.gradio['gptq_for_llama_info'] = gr.Markdown('Legacy loader for compatibility with older GPUs. ExLlamav2_HF or AutoGPTQ are preferred for GPTQ models when supported.')
                             shared.gradio['exllamav2_info'] = gr.Markdown("ExLlamav2_HF is recommended over ExLlamav2 for better integration with extensions and more consistent sampling behavior across loaders.")
-                            shared.gradio['llamacpp_HF_info'] = gr.Markdown("llamacpp_HF loads llama.cpp as a Transformers model. To use it, download a tokenizer in HF format for your GGUF:\n\n1. Create a folder inside models/\n2. Place your GGUF in the new folder.\n3. Add the original model's tokenizer files there: `tokenizer.model`, `tokenizer_config.json`, `tokenizer.json`, and `special_tokens_map.json`.")
+                            shared.gradio['llamacpp_HF_info'] = gr.Markdown("llamacpp_HF loads llama.cpp as a Transformers model. To use it, you need to place your GGUF in a subfolder of models/ with the necessary tokenizer files.\n\nYou can use the \"llamacpp_HF creator\" menu to do that automatically.")
 
             with gr.Column():
                 with gr.Row():
                     shared.gradio['autoload_model'] = gr.Checkbox(value=shared.settings['autoload_model'], label='Autoload the model', info='Whether to load the model as soon as it is selected in the Model dropdown.', interactive=not mu)
 
-                shared.gradio['custom_model_menu'] = gr.Textbox(label="Download model or LoRA", info="Enter the Hugging Face username/model path, for instance: facebook/galactica-125m. To specify a branch, add it at the end after a \":\" character like this: facebook/galactica-125m:main. To download a single file, enter its name in the second box.", interactive=not mu)
-                shared.gradio['download_specific_file'] = gr.Textbox(placeholder="File name (for GGUF models)", show_label=False, max_lines=1, interactive=not mu)
-                with gr.Row():
-                    shared.gradio['download_model_button'] = gr.Button("Download", variant='primary', interactive=not mu)
-                    shared.gradio['get_file_list'] = gr.Button("Get file list", interactive=not mu)
+                with gr.Tab("Download"):
+                    shared.gradio['custom_model_menu'] = gr.Textbox(label="Download model or LoRA", info="Enter the Hugging Face username/model path, for instance: facebook/galactica-125m. To specify a branch, add it at the end after a \":\" character like this: facebook/galactica-125m:main. To download a single file, enter its name in the second box.", interactive=not mu)
+                    shared.gradio['download_specific_file'] = gr.Textbox(placeholder="File name (for GGUF models)", show_label=False, max_lines=1, interactive=not mu)
+                    with gr.Row():
+                        shared.gradio['download_model_button'] = gr.Button("Download", variant='primary', interactive=not mu)
+                        shared.gradio['get_file_list'] = gr.Button("Get file list", interactive=not mu)
+
+                with gr.Tab("llamacpp_HF creator"):
+                    with gr.Row():
+                        shared.gradio['gguf_menu'] = gr.Dropdown(choices=utils.get_available_ggufs(), value=lambda: shared.model_name, label='Choose your GGUF', elem_classes='slim-dropdown', interactive=not mu)
+                        ui.create_refresh_button(shared.gradio['gguf_menu'], lambda: None, lambda: {'choices': utils.get_available_ggufs()}, 'refresh-button', interactive=not mu)
+
+                    shared.gradio['unquantized_url'] = gr.Textbox(label="Enter the URL for the original (unquantized) model", info="Example: https://huggingface.co/lmsys/vicuna-13b-v1.5", max_lines=1)
+                    shared.gradio['create_llamacpp_hf_button'] = gr.Button("Submit", variant="primary", interactive=not mu)
+                    gr.Markdown("This will move your gguf file into a subfolder of `models` along with the necessary tokenizer files.")
+
+                with gr.Tab("Customize instruction template"):
+                    with gr.Row():
+                        shared.gradio['customized_template'] = gr.Dropdown(choices=utils.get_available_instruction_templates(), value='None', label='Select the desired instruction template', elem_classes='slim-dropdown')
+                        ui.create_refresh_button(shared.gradio['customized_template'], lambda: None, lambda: {'choices': utils.get_available_instruction_templates()}, 'refresh-button', interactive=not mu)
+
+                    shared.gradio['customized_template_submit'] = gr.Button("Submit", variant="primary", interactive=not mu)
+                    gr.Markdown("This allows you to set a customized template for the model currently selected in the \"Model loader\" menu. Whenever the model gets loaded, this template will be used in place of the template specified in the model's medatada, which sometimes is wrong.")
 
                 with gr.Row():
                     shared.gradio['model_status'] = gr.Markdown('No model is loaded' if shared.model_name == 'None' else 'Ready')
@@ -203,6 +223,8 @@ def create_event_handlers():
     shared.gradio['download_model_button'].click(download_model_wrapper, gradio('custom_model_menu', 'download_specific_file'), gradio('model_status'), show_progress=True)
     shared.gradio['get_file_list'].click(partial(download_model_wrapper, return_links=True), gradio('custom_model_menu', 'download_specific_file'), gradio('model_status'), show_progress=True)
     shared.gradio['autoload_model'].change(lambda x: gr.update(visible=not x), gradio('autoload_model'), gradio('load_model'))
+    shared.gradio['create_llamacpp_hf_button'].click(create_llamacpp_hf, gradio('gguf_menu', 'unquantized_url'), gradio('model_status'), show_progress=True)
+    shared.gradio['customized_template_submit'].click(save_instruction_template, gradio('model_menu', 'customized_template'), gradio('model_status'), show_progress=True)
 
 
 def load_model_wrapper(selected_model, loader, autoload=False):
@@ -244,27 +266,58 @@ def load_lora_wrapper(selected_loras):
 
 def download_model_wrapper(repo_id, specific_file, progress=gr.Progress(), return_links=False, check=False):
     try:
-        progress(0.0)
         downloader = importlib.import_module("download-model").ModelDownloader()
+
+        progress(0.0)
         model, branch = downloader.sanitize_model_and_branch_names(repo_id, None)
+
         yield ("Getting the download links from Hugging Face")
         links, sha256, is_lora, is_llamacpp = downloader.get_download_links_from_huggingface(model, branch, text_only=False, specific_file=specific_file)
         if return_links:
-            yield '\n\n'.join([f"`{Path(link).name}`" for link in links])
+            output = "```\n"
+            for link in links:
+                output += f"{Path(link).name}" + "\n"
+
+            output += "```"
+            yield output
             return
 
         yield ("Getting the output folder")
-        base_folder = shared.args.lora_dir if is_lora else shared.args.model_dir
-        output_folder = downloader.get_output_folder(model, branch, is_lora, is_llamacpp=is_llamacpp, base_folder=base_folder)
+        output_folder = downloader.get_output_folder(model, branch, is_lora, is_llamacpp=is_llamacpp)
         if check:
             progress(0.5)
+
             yield ("Checking previously downloaded files")
             downloader.check_model_files(model, branch, links, sha256, output_folder)
             progress(1.0)
         else:
             yield (f"Downloading file{'s' if len(links) > 1 else ''} to `{output_folder}/`")
             downloader.download_model_files(model, branch, links, sha256, output_folder, progress_bar=progress, threads=4, is_llamacpp=is_llamacpp)
-            yield ("Done!")
+
+            yield (f"Model successfully saved to `{output_folder}/`.")
+    except:
+        progress(1.0)
+        yield traceback.format_exc().replace('\n', '\n\n')
+
+
+def create_llamacpp_hf(gguf_name, unquantized_url, progress=gr.Progress()):
+    try:
+        downloader = importlib.import_module("download-model").ModelDownloader()
+
+        progress(0.0)
+        model, branch = downloader.sanitize_model_and_branch_names(unquantized_url, None)
+
+        yield ("Getting the tokenizer files links from Hugging Face")
+        links, sha256, is_lora, is_llamacpp = downloader.get_download_links_from_huggingface(model, branch, text_only=True)
+        output_folder = Path(shared.args.model_dir) / (re.sub(r'(?i)\.gguf$', '', gguf_name) + "-HF")
+
+        yield (f"Downloading tokenizer to `{output_folder}`")
+        downloader.download_model_files(model, branch, links, sha256, output_folder, progress_bar=progress, threads=4, is_llamacpp=False)
+
+        # Move the GGUF
+        (Path(shared.args.model_dir) / gguf_name).rename(output_folder / gguf_name)
+
+        yield (f"Model saved to `{output_folder}/`.\n\nYou can now load it using llamacpp_HF.")
     except:
         progress(1.0)
         yield traceback.format_exc().replace('\n', '\n\n')
diff --git a/modules/utils.py b/modules/utils.py
index de6d32dc..4b65736b 100644
--- a/modules/utils.py
+++ b/modules/utils.py
@@ -76,7 +76,16 @@ def get_available_models():
     model_list = []
     for item in list(Path(f'{shared.args.model_dir}/').glob('*')):
         if not item.name.endswith(('.txt', '-np', '.pt', '.json', '.yaml', '.py')) and 'llama-tokenizer' not in item.name:
-            model_list.append(re.sub('.pth$', '', item.name))
+            model_list.append(item.name)
+
+    return ['None'] + sorted(model_list, key=natural_keys)
+
+
+def get_available_ggufs():
+    model_list = []
+    for item in Path(f'{shared.args.model_dir}/').glob('*'):
+        if item.is_file() and item.name.lower().endswith(".gguf"):
+            model_list.append(item.name)
 
     return ['None'] + sorted(model_list, key=natural_keys)
 
@@ -105,7 +114,7 @@ def get_available_instruction_templates():
     if os.path.exists(path):
         paths = (x for x in Path(path).iterdir() if x.suffix in ('.json', '.yaml', '.yml'))
 
-    return ['Select template to load...'] + sorted(set((k.stem for k in paths)), key=natural_keys)
+    return ['None'] + sorted(set((k.stem for k in paths)), key=natural_keys)
 
 
 def get_available_extensions():
diff --git a/requirements.txt b/requirements.txt
index 3a16e1ef..bdbdf819 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -24,37 +24,37 @@ tqdm
 wandb
 
 # bitsandbytes
-bitsandbytes==0.41.1; platform_system != "Windows"
+bitsandbytes==0.42.*; platform_system != "Windows"
 https://github.com/jllllll/bitsandbytes-windows-webui/releases/download/wheels/bitsandbytes-0.41.1-py3-none-win_amd64.whl; platform_system == "Windows"
 
 # llama-cpp-python (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx2-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx2-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx2-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx2-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, no tensor cores)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.42+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.42+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.42+cu121-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.42+cu121-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.43+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.43+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.43+cu121-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.43+cu121-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, tensor cores)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.42+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.42+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.42+cu121-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.42+cu121-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.43+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.43+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.43+cu121-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.43+cu121-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # CUDA wheels
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/jllllll/flash-attention/releases/download/v2.3.4/flash_attn-2.3.4+cu121torch2.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/jllllll/flash-attention/releases/download/v2.3.4/flash_attn-2.3.4+cu121torch2.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.3.4/flash_attn-2.3.4+cu122torch2.1cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements_amd.txt b/requirements_amd.txt
index 21cb90d6..8e11d3c5 100644
--- a/requirements_amd.txt
+++ b/requirements_amd.txt
@@ -28,18 +28,18 @@ bitsandbytes==0.38.1; platform_system != "Windows"
 https://github.com/jllllll/bitsandbytes-windows-webui/releases/download/wheels/bitsandbytes-0.38.1-py3-none-win_amd64.whl; platform_system == "Windows"
 
 # llama-cpp-python (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx2-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx2-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx2-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx2-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.42+rocm5.6.1-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.42+rocm5.6.1-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.43+rocm5.6.1-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.43+rocm5.6.1-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
 https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_llama-0.1.1+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_llama-0.1.1+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
diff --git a/requirements_amd_noavx2.txt b/requirements_amd_noavx2.txt
index 349acc31..89edbb07 100644
--- a/requirements_amd_noavx2.txt
+++ b/requirements_amd_noavx2.txt
@@ -28,16 +28,16 @@ bitsandbytes==0.38.1; platform_system != "Windows"
 https://github.com/jllllll/bitsandbytes-windows-webui/releases/download/wheels/bitsandbytes-0.38.1-py3-none-win_amd64.whl; platform_system == "Windows"
 
 # llama-cpp-python (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # AMD wheels
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
 https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_llama-0.1.1+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_llama-0.1.1+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
diff --git a/requirements_apple_intel.txt b/requirements_apple_intel.txt
index 94996ec1..4030bf9a 100644
--- a/requirements_apple_intel.txt
+++ b/requirements_apple_intel.txt
@@ -24,14 +24,14 @@ tqdm
 wandb
 
 # bitsandbytes
-bitsandbytes==0.41.1; platform_system != "Windows"
+bitsandbytes==0.42.*; platform_system != "Windows"
 https://github.com/jllllll/bitsandbytes-windows-webui/releases/download/wheels/bitsandbytes-0.41.1-py3-none-win_amd64.whl; platform_system == "Windows"
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.42-cp311-cp311-macosx_11_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "20.0.0" and platform_release < "21.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.42-cp310-cp310-macosx_11_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "20.0.0" and platform_release < "21.0.0" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.42-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.42-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.42-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.42-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1-py3-none-any.whl
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.43-cp311-cp311-macosx_11_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "20.0.0" and platform_release < "21.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.43-cp310-cp310-macosx_11_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "20.0.0" and platform_release < "21.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.43-cp311-cp311-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.43-cp310-cp310-macosx_12_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.43-cp311-cp311-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.43-cp310-cp310-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2-py3-none-any.whl
diff --git a/requirements_apple_silicon.txt b/requirements_apple_silicon.txt
index 3aa9391b..8cea6d62 100644
--- a/requirements_apple_silicon.txt
+++ b/requirements_apple_silicon.txt
@@ -24,16 +24,16 @@ tqdm
 wandb
 
 # bitsandbytes
-bitsandbytes==0.41.1; platform_system != "Windows"
+bitsandbytes==0.42.*; platform_system != "Windows"
 https://github.com/jllllll/bitsandbytes-windows-webui/releases/download/wheels/bitsandbytes-0.41.1-py3-none-win_amd64.whl; platform_system == "Windows"
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.42-cp311-cp311-macosx_11_0_arm64.whl; platform_system == "Darwin" and platform_release >= "20.0.0" and platform_release < "21.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.42-cp310-cp310-macosx_11_0_arm64.whl; platform_system == "Darwin" and platform_release >= "20.0.0" and platform_release < "21.0.0" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.42-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.42-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.42-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.42-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.42-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.42-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1-py3-none-any.whl
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.43-cp311-cp311-macosx_11_0_arm64.whl; platform_system == "Darwin" and platform_release >= "20.0.0" and platform_release < "21.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.43-cp310-cp310-macosx_11_0_arm64.whl; platform_system == "Darwin" and platform_release >= "20.0.0" and platform_release < "21.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.43-cp311-cp311-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.43-cp310-cp310-macosx_12_0_arm64.whl; platform_system == "Darwin" and platform_release >= "21.0.0" and platform_release < "22.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.43-cp311-cp311-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.43-cp310-cp310-macosx_13_0_arm64.whl; platform_system == "Darwin" and platform_release >= "22.0.0" and platform_release < "23.0.0" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.43-cp311-cp311-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/metal/llama_cpp_python-0.2.43-cp310-cp310-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2-py3-none-any.whl
diff --git a/requirements_cpu_only.txt b/requirements_cpu_only.txt
index 16aa1676..283e2b19 100644
--- a/requirements_cpu_only.txt
+++ b/requirements_cpu_only.txt
@@ -24,11 +24,11 @@ tqdm
 wandb
 
 # bitsandbytes
-bitsandbytes==0.41.1; platform_system != "Windows"
+bitsandbytes==0.42.*; platform_system != "Windows"
 https://github.com/jllllll/bitsandbytes-windows-webui/releases/download/wheels/bitsandbytes-0.41.1-py3-none-win_amd64.whl; platform_system == "Windows"
 
 # llama-cpp-python (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx2-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx2-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx2-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx2-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx2-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx2-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
diff --git a/requirements_cpu_only_noavx2.txt b/requirements_cpu_only_noavx2.txt
index 1fa23bcf..ba2d3d10 100644
--- a/requirements_cpu_only_noavx2.txt
+++ b/requirements_cpu_only_noavx2.txt
@@ -24,11 +24,11 @@ tqdm
 wandb
 
 # bitsandbytes
-bitsandbytes==0.41.1; platform_system != "Windows"
+bitsandbytes==0.42.*; platform_system != "Windows"
 https://github.com/jllllll/bitsandbytes-windows-webui/releases/download/wheels/bitsandbytes-0.41.1-py3-none-win_amd64.whl; platform_system == "Windows"
 
 # llama-cpp-python (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
diff --git a/requirements_noavx2.txt b/requirements_noavx2.txt
index 4d9caf36..5025fe9c 100644
--- a/requirements_noavx2.txt
+++ b/requirements_noavx2.txt
@@ -24,37 +24,37 @@ tqdm
 wandb
 
 # bitsandbytes
-bitsandbytes==0.41.1; platform_system != "Windows"
+bitsandbytes==0.42.*; platform_system != "Windows"
 https://github.com/jllllll/bitsandbytes-windows-webui/releases/download/wheels/bitsandbytes-0.41.1-py3-none-win_amd64.whl; platform_system == "Windows"
 
 # llama-cpp-python (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.42+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cpu/llama_cpp_python-0.2.43+cpuavx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, no tensor cores)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.42+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.42+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.42+cu121avx-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.42+cu121avx-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.43+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.43+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.43+cu121avx-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda-0.2.43+cu121avx-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # llama-cpp-python (CUDA, tensor cores)
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.42+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.42+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.42+cu121avx-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.42+cu121avx-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.43+cu121avx-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.43+cu121avx-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.43+cu121avx-cp311-cp311-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/textgen-webui/llama_cpp_python_cuda_tensorcores-0.2.43+cu121avx-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 
 # CUDA wheels
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.1/exllamav2-0.0.13.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.13.2/exllamav2-0.0.13.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/jllllll/flash-attention/releases/download/v2.3.4/flash_attn-2.3.4+cu121torch2.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/jllllll/flash-attention/releases/download/v2.3.4/flash_attn-2.3.4+cu121torch2.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.3.4/flash_attn-2.3.4+cu122torch2.1cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
diff --git a/requirements_nowheels.txt b/requirements_nowheels.txt
index 5bf2cc4c..2b6fa38c 100644
--- a/requirements_nowheels.txt
+++ b/requirements_nowheels.txt
@@ -24,5 +24,5 @@ tqdm
 wandb
 
 # bitsandbytes
-bitsandbytes==0.41.1; platform_system != "Windows"
+bitsandbytes==0.42.*; platform_system != "Windows"
 https://github.com/jllllll/bitsandbytes-windows-webui/releases/download/wheels/bitsandbytes-0.41.1-py3-none-win_amd64.whl; platform_system == "Windows"