Merge pull request #206 from oobabooga/llama-4bit

Add LLaMA 4-bit support

README.md

@@ -138,6 +138,7 @@ 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. |
 | `--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. |

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 (shared.args.cpu or shared.args.load_in_8bit or shared.args.load_in_4bit 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 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 +87,28 @@ def load_model(model_name):
 
         return model, tokenizer
 
+    # 4-bit LLaMA
+    elif shared.args.load_in_4bit:
+        sys.path.insert(0, os.path.abspath(Path("repositories/GPTQ-for-LLaMa")))
+
+        from llama import load_quant
+
+        path_to_model = Path(f'models/{model_name}')
+        pt_model = ''
+        if path_to_model.name.lower().startswith('llama-7b'):
+            pt_model = 'llama-7b-4bit.pt'
+        if path_to_model.name.lower().startswith('llama-13b'):
+            pt_model = 'llama-13b-4bit.pt'
+        if path_to_model.name.lower().startswith('llama-30b'):
+            pt_model = 'llama-30b-4bit.pt'
+
+        if not Path(f"models/{pt_model}").exists():
+            print(f"Could not find models/{pt_model}, exiting...")
+            exit()
+
+        model = load_quant(path_to_model, Path(f"models/{pt_model}"), 4)
+        model = model.to(torch.device('cuda:0'))
+
     # Custom
     else:
         command = "AutoModelForCausalLM.from_pretrained"

modules/shared.py

@@ -68,6 +68,7 @@ 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('--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.')