From 31c297d7e04e02f1aeb92215fea69dbd06101404 Mon Sep 17 00:00:00 2001
From: oobabooga <112222186+oobabooga@users.noreply.github.com>
Date: Tue, 4 Jul 2023 18:50:01 -0700
Subject: [PATCH] Various changes

---
 characters/instruction-following/Xgen | 4 ----
 models/config.yaml                    | 4 ++--
 modules/models.py                     | 2 --
 3 files changed, 2 insertions(+), 8 deletions(-)
 delete mode 100644 characters/instruction-following/Xgen

diff --git a/characters/instruction-following/Xgen b/characters/instruction-following/Xgen
deleted file mode 100644
index c7b76d32..00000000
--- a/characters/instruction-following/Xgen
+++ /dev/null
@@ -1,4 +0,0 @@
-user: "### Human:"
-bot: "### Assistant:"
-turn_template: "<|user|><|user-message|>\n<|bot|><|bot-message|>\n"
-context: "A chat between a curious human and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the human's questions.\n\n"
diff --git a/models/config.yaml b/models/config.yaml
index 9e81def1..d81eac94 100644
--- a/models/config.yaml
+++ b/models/config.yaml
@@ -240,6 +240,6 @@ TheBloke_WizardLM-30B-GPTQ:
   truncation_length: 8192
 .*superhot-8k:
   truncation_length: 8192
-.*xgen:
+.*xgen.*-inst:
   truncation_length: 8192
-  instruction_template: 'XGen'
+  instruction_template: 'Vicuna-v0'
diff --git a/modules/models.py b/modules/models.py
index d6a3585f..f12e700c 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -94,8 +94,6 @@ def load_tokenizer(model_name, model):
     if any(s in model_name.lower() for s in ['gpt-4chan', 'gpt4chan']) and Path(f"{shared.args.model_dir}/gpt-j-6B/").exists():
         tokenizer = AutoTokenizer.from_pretrained(Path(f"{shared.args.model_dir}/gpt-j-6B/"))
     elif model.__class__.__name__ in ['LlamaForCausalLM', 'LlamaGPTQForCausalLM', 'ExllamaHF']:
-        if any(s in model_name.lower() for s in ['xgen']) and Path(f"{shared.args.model_dir}/{model_name}/").exists():
-            tokenizer = AutoTokenizer.from_pretrained(Path(f"{shared.args.model_dir}/{model_name}/"), trust_remote_code=shared.args.trust_remote_code)
         # Try to load an universal LLaMA tokenizer
         if not any(s in shared.model_name.lower() for s in ['llava', 'oasst']):
             for p in [Path(f"{shared.args.model_dir}/llama-tokenizer/"), Path(f"{shared.args.model_dir}/oobabooga_llama-tokenizer/")]: