Co

.github/workflows/build_on_pr.yml

@@ -208,7 +208,7 @@ jobs:
 
       - name: Execute Unit Testing
         run: |
-          CURL_CA_BUNDLE="" PYTHONPATH=$PWD pytest --testmon --testmon-cov=. tests/
+          CURL_CA_BUNDLE="" PYTHONPATH=$PWD pytest --testmon --testmon-cov=. --durations=10 tests/
         env:
           DATA: /data/scratch/cifar-10
           NCCL_SHM_DISABLE: 1

.github/workflows/build_on_schedule.yml

@@ -3,7 +3,7 @@ name: Build on Schedule
 on:
   schedule:
     # run at 00:00 of every Sunday
-    - cron:  '0 0 * * *'
+    - cron: "0 0 * * *"
   workflow_dispatch:
 
 jobs:
@@ -60,7 +60,7 @@ jobs:
       - name: Unit Testing
         if: steps.check-avai.outputs.avai == 'true'
         run: |
-          PYTHONPATH=$PWD pytest tests
+          PYTHONPATH=$PWD pytest --durations=0 tests
         env:
           DATA: /data/scratch/cifar-10
           LD_LIBRARY_PATH: /github/home/.tensornvme/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64

.github/workflows/run_chatgpt_examples.yml

@@ -4,11 +4,10 @@ on:
   pull_request:
     types: [synchronize, opened, reopened]
     paths:
-      - 'applications/Chat/coati/**'
-      - 'applications/Chat/requirements.txt'
-      - 'applications/Chat/setup.py'
-      - 'applications/Chat/examples/**'
-
+      - "applications/Chat/coati/**"
+      - "applications/Chat/requirements.txt"
+      - "applications/Chat/setup.py"
+      - "applications/Chat/examples/**"
 
 jobs:
   tests:
@@ -38,10 +37,7 @@ jobs:
 
       - name: Install Transformers
         run: |
-          cd applications/Chat
-          git clone https://github.com/hpcaitech/transformers
-          cd transformers
-          pip install -v .
+          pip install transformers==4.30.2
 
       - name: Execute Examples
         run: |

applications/Chat/README.md

@@ -98,12 +98,9 @@ pip install .
 ```
 
 ### Install the Transformers
-Given Hugging Face hasn't officially supported the LLaMA models, We fork a branch of Transformers that can be compatible with our code
 
 ```shell
-git clone https://github.com/hpcaitech/transformers
-cd transformers
-pip install .
+pip install transformers==4.30.2
 ```
 
 ## How to use?

applications/Chat/coati/ray/utils.py

@@ -9,10 +9,8 @@
 from coati.models.gpt import GPTRM, GPTActor, GPTCritic
 from coati.models.llama import LlamaActor, LlamaCritic, LlamaRM
 from coati.models.opt import OPTRM, OPTActor, OPTCritic
-from coati.models.roberta import RoBERTaActor, RoBERTaCritic, RoBERTaRM
 from coati.trainer.strategies import DDPStrategy, GeminiStrategy, LowLevelZeroStrategy
-from coati.utils import prepare_llama_tokenizer_and_embedding
-from transformers import AutoTokenizer, BloomTokenizerFast, GPT2Tokenizer, LlamaTokenizer, RobertaTokenizer
+from transformers import AutoTokenizer, BloomTokenizerFast, GPT2Tokenizer, LlamaTokenizer
 
 
 def is_rank_0() -> bool:
@@ -36,8 +34,6 @@ def get_actor_from_args(model: str, pretrained: str = None, config=None, lora_ra
         actor = OPTActor(pretrained=pretrained, config=config, lora_rank=lora_rank)
     elif model == 'llama':
         actor = LlamaActor(pretrained=pretrained, config=config, lora_rank=lora_rank)
-    elif model == 'roberta':
-        actor = RoBERTaActor(pretrained=pretrained, config=config, lora_rank=lora_rank)
     else:
         raise ValueError(f'Unsupported actor model "{model}"')
     return actor
@@ -52,8 +48,6 @@ def get_critic_from_args(model: str, pretrained: str = None, config=None, lora_r
         critic = OPTCritic(pretrained=pretrained, lora_rank=lora_rank, config=config, use_action_mask=True)
     elif model == 'llama':
         critic = LlamaCritic(pretrained=pretrained, lora_rank=lora_rank, config=config, use_action_mask=True)
-    elif model == 'roberta':
-        critic = RoBERTaCritic(pretrained=pretrained, lora_rank=lora_rank, config=config, use_action_mask=True)
     else:
         raise ValueError(f'Unsupported reward model "{model}"')
     return critic
@@ -68,8 +62,6 @@ def get_reward_model_from_args(model: str, pretrained: str = None, config=None):
         reward_model = OPTRM(pretrained=pretrained, config=config)
     elif model == 'llama':
         reward_model = LlamaRM(pretrained=pretrained, config=config)
-    elif model == 'roberta':
-        reward_model = RoBERTaRM(pretrained=pretrained, config=config)
     else:
         raise ValueError(f'Unsupported reward model "{model}"')
     return reward_model
@@ -101,8 +93,6 @@ def get_tokenizer_from_args(model: str, **kwargs):
     elif model == 'llama':
         pretrain_path = kwargs["pretrain"]
         tokenizer = AutoTokenizer.from_pretrained(pretrain_path)
-    elif model == 'roberta':
-        tokenizer = RobertaTokenizer.from_pretrained("roberta-base")
     else:
         raise ValueError(f'Unsupported model "{model}"')