Do not impose instruct mode while loading models

models/config.yaml

@@ -49,200 +49,142 @@ llama-65b-gptq-3bit:
 .*(gr1024|1024g|groupsize1024):
   groupsize: 1024
 .*(oasst|openassistant-|stablelm-7b-sft-v7-epoch-3):
-  mode: 'instruct'
   instruction_template: 'Open Assistant'
   skip_special_tokens: false
 (?!.*galactica)(?!.*reward).*openassistant:
-  mode: 'instruct'
   instruction_template: 'Open Assistant'
   skip_special_tokens: false
 (?!.*v0)(?!.*1.1)(?!.*1_1)(?!.*stable)(?!.*chinese).*vicuna:
-  mode: 'instruct'
   instruction_template: 'Vicuna-v0'
 .*vicuna.*v0:
-  mode: 'instruct'
   instruction_template: 'Vicuna-v0'
 .*vicuna.*(1.1|1_1|1.3|1_3):
-  mode: 'instruct'
   instruction_template: 'Vicuna-v1.1'
 .*vicuna.*(1.5|1_5):
-  mode: 'instruct'
   instruction_template: 'Vicuna-v1.1'
   truncation_length: 4096
   rms_norm_eps: 5.0e-6  
 .*stable.*vicuna:
-  mode: 'instruct'
   instruction_template: 'StableVicuna'
 (?!.*chat).*chinese-vicuna:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
 .*chinese-vicuna.*chat:
-  mode: 'instruct'
   instruction_template: 'Chinese-Vicuna-Chat'
 .*alpaca:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
 .*alpaca-native-4bit:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
   wbits: 4
   groupsize: 128
 .*galactica:
   skip_special_tokens: false
 .*dolly-v[0-9]-[0-9]*b:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
   skip_special_tokens: false
   custom_stopping_strings: '"### End"'
 .*koala:
-  mode: 'instruct'
   instruction_template: 'Koala'
 .*chatglm:
-  mode: 'instruct'
   instruction_template: 'ChatGLM'
 .*metharme:
-  mode: 'instruct'
   instruction_template: 'Metharme'
 .*llava:
-  mode: 'instruct'
   model_type: 'llama'
   instruction_template: 'LLaVA'
   custom_stopping_strings: '"\n###"'
 .*raven:
-  mode: 'instruct'
   instruction_template: 'RWKV-Raven'
 .*ctx8192:
   truncation_length: 8192
 .*moss-moon.*sft:
-  mode: 'instruct'
   instruction_template: 'MOSS'
 .*stablelm-tuned:
-  mode: 'instruct'
   instruction_template: 'StableLM'
   truncation_length: 4096
 .*stablelm-base:
   truncation_length: 4096
 .*galactica.*finetuned:
-  mode: 'instruct'
   instruction_template: 'Galactica Finetuned'
 .*galactica.*-v2:
-  mode: 'instruct'
   instruction_template: 'Galactica v2'
 (?!.*finetuned)(?!.*-v2).*galactica:
-  mode: 'instruct'
   instruction_template: 'Galactica'
 .*guanaco:
-  mode: 'instruct'
   instruction_template: 'Guanaco non-chat'
 .*baize:
-  mode: 'instruct'
   instruction_template: 'Baize'
 .*mpt-.*instruct:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
 .*mpt-.*chat:
-  mode: 'instruct'
   instruction_template: 'MPT-Chat'
 (?!.*-flan-)(?!.*-t5-).*lamini-:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
 .*incite.*chat:
-  mode: 'instruct'
   instruction_template: 'INCITE-Chat'
 .*incite.*instruct:
-  mode: 'instruct'
   instruction_template: 'INCITE-Instruct'
 .*wizard.*mega:
-  mode: 'instruct'
   instruction_template: 'Wizard-Mega'
   custom_stopping_strings: '"</s>"'
 .*ziya-:
-  mode: 'instruct'
   instruction_template: 'Ziya'
 .*koalpaca:
-  mode: 'instruct'
   instruction_template: 'KoAlpaca'
 .*openbuddy:
-  mode: 'instruct'
   instruction_template: 'OpenBuddy'
 (?!.*chat).*vigogne:
-  mode: 'instruct'
   instruction_template: 'Vigogne-Instruct'
 .*vigogne.*chat:
-  mode: 'instruct'
   instruction_template: 'Vigogne-Chat'
 .*(llama-deus|supercot|llama-natural-instructions|open-llama-0.3t-7b-instruct-dolly-hhrlhf|open-llama-0.3t-7b-open-instruct):
-  mode: 'instruct'
   instruction_template: 'Alpaca'
 .*bactrian:
-  mode: 'instruct'
   instruction_template: 'Bactrian'
 .*(h2ogpt-oig-|h2ogpt-oasst1-|h2ogpt-research-oasst1-):
-  mode: 'instruct'
   instruction_template: 'H2O-human_bot'
 .*h2ogpt-gm-:
-  mode: 'instruct'
   instruction_template: 'H2O-prompt_answer'
 .*manticore:
-  mode: 'instruct'
   instruction_template: 'Manticore Chat'
 .*bluemoonrp-(30|13)b:
-  mode: 'instruct'
   instruction_template: 'Bluemoon'
   truncation_length: 4096
 .*Nous-Hermes-13b:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
 .*airoboros:
-  mode: 'instruct'
   instruction_template: 'Vicuna-v1.1'
 .*airoboros.*1.2:
-  mode: 'instruct'
   instruction_template: 'Airoboros-v1.2'
 .*alpa(cino|sta):
-  mode: 'instruct'
   instruction_template: 'Alpaca'
 .*hippogriff:
-  mode: 'instruct'
   instruction_template: 'Hippogriff'
 .*lazarus:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
 .*guanaco-.*(7|13|33|65)b:
-  mode: 'instruct'
   instruction_template: 'Guanaco'
 .*hypermantis:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
 .*open-llama-.*-open-instruct:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
 .*starcoder-gpteacher-code-instruct:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
 .*tulu:
-  mode: 'instruct'
   instruction_template: 'Tulu'
 .*chronos:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
 .*samantha:
-  mode: 'instruct'
   instruction_template: 'Samantha'
 .*wizardcoder:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
 .*starchat-beta:
-  mode: 'instruct'
   instruction_template: 'Starchat-Beta'
   custom_stopping_strings: '"<|end|>"'
 .*minotaur:
-  mode: 'instruct'
   instruction_template: 'Minotaur'
 .*minotaur-15b:
   truncation_length: 8192
 .*orca_mini:
-  mode: 'instruct'
   instruction_template: 'Orca Mini'
 .*landmark:
   truncation_length: 8192
@@ -252,56 +194,42 @@ llama-65b-gptq-3bit:
   truncation_length: 8192
   instruction_template: 'Vicuna-v0'
 .*(platypus|gplatty|superplatty):
-  mode: 'instruct'
   instruction_template: 'Alpaca'
 .*longchat:
-  mode: 'instruct'
   instruction_template: 'Vicuna-v1.1'
 .*vicuna-33b:
-  mode: 'instruct'
   instruction_template: 'Vicuna-v1.1'
 .*redmond-hermes-coder:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
   truncation_length: 8192
 .*wizardcoder-15b:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
   truncation_length: 8192
 .*wizardlm:
-  mode: 'instruct'
   instruction_template: 'Vicuna-v1.1'
 .*godzilla:
-  mode: 'instruct'
   instruction_template: 'Alpaca'
 .*llama-(2|v2):
   truncation_length: 4096
   rms_norm_eps: 5.0e-6
 .*llama-(2|v2).*chat:
-  mode: 'instruct'
   instruction_template: 'Llama-v2'
 .*70b.*ggml.*\.bin:
   n_gqa: 8
 .*newhope:
-  mode: 'instruct'
   instruction_template: 'NewHope'
 .*stablebeluga2:
-  mode: 'instruct'
   instruction_template: 'StableBeluga2'
   truncation_length: 4096
   rms_norm_eps: 5.0e-6
 .*openchat:
-  mode: 'instruct'
   instruction_template: 'OpenChat'
 .*falcon.*-instruct:
-  mode: 'instruct'
 .*(openorca-platypus2):
-  mode: 'instruct'
   instruction_template: 'OpenOrca-Platypus2'
   custom_stopping_strings: '"### Instruction:", "### Response:"'
   rms_norm_eps: 5.0e-6
 .*codellama:
   rope_freq_base: 1000000
 .*codellama.*instruct:
-  mode: 'instruct'
   instruction_template: 'Llama-v2'

modules/ui_model_menu.py

@@ -15,6 +15,7 @@ from modules.LoRA import add_lora_to_model
 from modules.models import load_model, unload_model
 from modules.models_settings import (
     apply_model_settings_to_state,
+    get_model_settings_from_yamls,
     save_model_settings,
     update_model_parameters
 )
@@ -181,23 +182,29 @@ def create_event_handlers():
 
 def load_model_wrapper(selected_model, loader, autoload=False):
     if not autoload:
-        yield f"The settings for {selected_model} have been updated.\nClick on \"Load\" to load it."
+        yield f"The settings for `{selected_model}` have been updated.\n\nClick on \"Load\" to load it."
         return
 
     if selected_model == 'None':
         yield "No model selected"
     else:
         try:
-            yield f"Loading {selected_model}..."
+            yield f"Loading `{selected_model}`..."
             shared.model_name = selected_model
             unload_model()
             if selected_model != '':
                 shared.model, shared.tokenizer = load_model(shared.model_name, loader)
 
             if shared.model is not None:
-                yield f"Successfully loaded {selected_model}"
+                output = f"Successfully loaded `{selected_model}`."
+
+                settings = get_model_settings_from_yamls(selected_model)
+                if 'instruction_template' in settings:
+                    output += '\n\nIt seems to be an instruction-following model with template "{}". In the chat tab, instruct or chat-instruct modes should be used.'.format(settings['instruction_template'])
+
+                yield output
             else:
-                yield f"Failed to load {selected_model}."
+                yield f"Failed to load `{selected_model}`."
         except:
             exc = traceback.format_exc()
             logger.error('Failed to load the model.')

server.py

@@ -174,7 +174,6 @@ if __name__ == "__main__":
         'model_type': 'None',
         'groupsize': 'None',
         'pre_layer': 0,
-        'mode': shared.settings['mode'],
         'skip_special_tokens': shared.settings['skip_special_tokens'],
         'custom_stopping_strings': shared.settings['custom_stopping_strings'],
         'truncation_length': shared.settings['truncation_length'],