add support for opimum bettertransformers

configs/gpt_neox_20b.yml

@@ -1,24 +1,25 @@
 base_model: EleutherAI/gpt-neox-20b
+base_model_config: EleutherAI/gpt-neox-20b
 base_model_ignore_patterns: pytorch*  # prefer safetensors
 model_type: GPTNeoXForCausalLM
 tokenizer_type: AutoTokenizer
-load_in_8bit: true
+load_in_8bit: false
+load_in_4bit: true
+load_4bit: false
 datasets:
-  - path: nomic-ai/gpt4all-j-prompt-generations
+  - path: vicgalle/alpaca-gpt4
     type: alpaca
-    shards: 4
-    shards_index: 0
 dataset_prepared_path: last_run_prepared
 val_set_size: 0.05
-adapter: lora
+adapter:
 lora_model_dir:
 sequence_len: 2048
 max_packed_sequence_len: 2048
-lora_r: 8
+lora_r: 64
 lora_alpha: 32
-lora_dropout: 0.05
+lora_dropout: 0.0
 lora_target_modules:
-  - query_key_value
+lora_target_linear: true
 lora_fan_in_fan_out: true  # pythia/GPTNeoX lora specific
 wandb_project: gpt4all-neox-20b
 wandb_watch:
@@ -26,14 +27,19 @@ wandb_run_id:
 wandb_log_model:
 output_dir: ./gpt4all-neox-20b
 gradient_accumulation_steps: 1
-micro_batch_size: 4
+micro_batch_size: 2
 num_epochs: 5
 learning_rate: 0.00003
-lr_scheduler: one_cycle
+optimizer: paged_adamw_32bit
+lr_scheduler: cosine
 train_on_inputs: false
 group_by_length: false
-bf16: True
-tf32: True
+bf16: false
+fp16: false
+float16: true
+tf32: true
+flash_optimum: true
 early_stopping_patience:
 resume_from_checkpoint:
 local_rank:
+gradient_checkpointing: true

requirements.txt

@@ -11,6 +11,7 @@ sentencepiece
 wandb
 einops
 xformers
+optimum
 # qlora things
 bert-score==0.3.13
 evaluate==0.4.0

scripts/finetune.py

@@ -6,6 +6,7 @@ import os
 import random
 import signal
 import sys
+from functools import partial
 from pathlib import Path
 from typing import Any, Dict, List, Optional, Union
 
@@ -19,6 +20,8 @@ from axolotl.utils.dict import DictDefault
 from axolotl.utils.models import load_model, load_tokenizer
 
 # add src to the pythonpath so we don't need to pip install this
+from optimum.bettertransformer import BetterTransformer
+
 from axolotl.utils.tokenization import check_dataset_labels
 from axolotl.utils.trainer import setup_trainer
 from axolotl.utils.validation import validate_config
@@ -264,12 +267,14 @@ def train(
 
     # In case we want to stop early with ctrl+c, this is a nice to have to save the pretrained model
     if cfg.local_rank == 0:
+        def terminate_handler(signum, frame, model):
+            if cfg.flash_optimum:
+                model = BetterTransformer.reverse(model)
+            model.save_pretrained(cfg.output_dir)
+            sys.exit(0)
         signal.signal(
             signal.SIGINT,
-            lambda signal, frame: (
-                model.save_pretrained(cfg.output_dir),
-                sys.exit(0),
-            ),
+            lambda signum, frame: terminate_handler(signum, frame, model)
         )
 
     logging.info("Starting trainer...")
@@ -299,6 +304,8 @@ def train(
     # TODO do we need this fix? https://huggingface.co/docs/accelerate/usage_guides/fsdp#saving-and-loading
     # only save on rank 0, otherwise it corrupts output on multi-GPU when multiple processes attempt to write the same file
     if cfg.local_rank == 0:
+        if cfg.flash_optimum:
+            model = BetterTransformer.reverse(model)
         model.save_pretrained(cfg.output_dir)
 
     # trainer.save_model(cfg.output_dir)  # TODO this may be needed for deepspeed to work? need to review another time

src/axolotl/utils/models.py

@@ -11,7 +11,8 @@ import bitsandbytes as bnb
 import torch
 import transformers
 from transformers import PreTrainedModel  # noqa: F401
-from transformers import (  # noqa: F401
+from optimum.bettertransformer import BetterTransformer
+from transformers import (
     AutoConfig,
     AutoModelForCausalLM,
     AutoTokenizer,
@@ -137,7 +138,7 @@ def load_model(
 
     if cfg.bf16:
         torch_dtype = torch.bfloat16
-    elif cfg.load_in_8bit or cfg.fp16:
+    elif cfg.load_in_8bit or cfg.fp16 or cfg.float16:
         torch_dtype = torch.float16
     else:
         torch_dtype = torch.float32
@@ -342,6 +343,9 @@ def load_model(
         logging.warning("there are no parameters that require gradient updates")
     model.config.use_cache = False
 
+    if cfg.flash_optimum:
+        model = BetterTransformer.transform(model)
+
     # TODO resume_from_checkpoint handling
     return model, lora_config
 

src/axolotl/utils/validation.py

@@ -57,6 +57,14 @@ def validate_config(cfg):
     if (cfg.base_model and "falcon" in cfg.base_model.lower()) and cfg.fsdp:
         raise ValueError("FSDP is not supported for falcon models")
 
+    if cfg.flash_optimum is True:
+        if cfg.adapter:
+            logging.warning("BetterTransformers probably doesn't work with PEFT adapters")
+        if cfg.fp16 or cfg.bf16:
+            raise ValueError("AMP is not supported with BetterTransformer")
+        if cfg.float16 is not True:
+            logging.warning("You should probably set float16 to true")
+
     # TODO
     # MPT 7b
     # https://github.com/facebookresearch/bitsandbytes/issues/25