add support for NCA

docs/config.qmd

@@ -138,7 +138,7 @@ test_datasets:
     data_files:
       - /workspace/data/eval.jsonl
 
-# use RL training: 'dpo', 'ipo', 'kto_pair'
+# use RL training: 'dpo', 'ipo', 'kto_pair', 'orpo', 'sppo_hard', 'nca_pair'
 rl:
 
 # Saves the desired chat template to the tokenizer_config.json for easier inferencing

requirements.txt

@@ -39,6 +39,6 @@ s3fs
 gcsfs
 # adlfs
 
-trl==0.8.5
+trl @ git+https://github.com/huggingface/trl.git@75de236c09bd5846f79c24d9bf371481b0b7582c
 zstandard==0.22.0
 fastcore

src/axolotl/cli/merge_lora.py

@@ -25,8 +25,6 @@ def do_cli(config: Path = Path("examples/"), **kwargs):
         load_in_8bit=False,
         load_in_4bit=False,
         flash_attention=False,
-        deepspeed=None,
-        fsdp=None,
         **kwargs,
     )
 

src/axolotl/core/trainer_builder.py

@@ -30,7 +30,7 @@ from transformers import (
 )
 from transformers.trainer_utils import seed_worker
 from transformers.utils import is_sagemaker_mp_enabled
-from trl import DPOTrainer, ORPOConfig, ORPOTrainer
+from trl import DPOConfig, DPOTrainer, ORPOConfig, ORPOTrainer
 from trl.trainer.utils import pad_to_length
 
 from axolotl.loraplus import create_loraplus_optimizer
@@ -1526,6 +1526,9 @@ class HFRLTrainerBuilder(TrainerBuilderBase):
         if self.cfg.rl == "orpo":
             training_args_cls = ORPOConfig
             training_args_kwargs["dataset_num_proc"] = self.cfg.dataset_processes
+        elif self.cfg.rl in ["dpo", "ipo", "kto_pair", "sppo_hard", "nca_pair"]:
+            training_args_cls = DPOConfig
+            training_args_kwargs["dataset_num_proc"] = self.cfg.dataset_processes
 
         training_args = training_args_cls(
             per_device_train_batch_size=self.cfg.micro_batch_size,
@@ -1550,8 +1553,8 @@ class HFRLTrainerBuilder(TrainerBuilderBase):
             dpo_trainer_kwargs["loss_type"] = "ipo"
             if self.cfg.dpo_label_smoothing:
                 dpo_trainer_kwargs["label_smoothing"] = self.cfg.dpo_label_smoothing
-        elif self.cfg.rl == "kto_pair":
+        elif self.cfg.rl in ["kto_pair", "sppo_hard", "nca_pair"]:
-            dpo_trainer_kwargs["loss_type"] = "kto_pair"
+            dpo_trainer_kwargs["loss_type"] = self.cfg.rl
         if self.eval_dataset:
             dpo_trainer_kwargs["eval_dataset"] = self.eval_dataset
         if self.cfg.adapter and self.peft_config:
@@ -1560,7 +1563,7 @@ class HFRLTrainerBuilder(TrainerBuilderBase):
             dpo_trainer_kwargs[
                 "precompute_ref_log_probs"
             ] = self.cfg.precompute_ref_log_probs
-        if self.cfg.rl in ["dpo", "ipo", "kto_pair"]:
+        if self.cfg.rl in ["dpo", "ipo", "kto_pair", "sppo_hard", "nca_pair"]:
             trainer_cls = AxolotlDPOTrainer
             dpo_trainer_kwargs["beta"] = self.cfg.dpo_beta or 0.1
             trainer_cls_args = [self.model, self.model_ref]

src/axolotl/prompt_strategies/dpo/mistral.py

@@ -0,0 +1,30 @@
+"""
+DPO strategies for mistral instruct
+"""
+
+
+def prompt_pairs(cfg):  # pylint: disable=possibly-unused-variable,unused-argument
+    def transform_fn(sample):
+        sample["prompt"] = f"[INST]{sample['prompt']}[/INST]"
+        sample["chosen"] = f"{sample['chosen']}"
+        sample["rejected"] = f"{sample['rejected']}"
+        return sample
+
+    return transform_fn
+
+
+def argilla_chat(
+    cfg,
+    **kwargs,
+):  # pylint: disable=possibly-unused-variable,unused-argument
+    """
+    for argilla/dpo-mix-7k conversations
+    """
+
+    def transform_fn(sample):
+        sample["prompt"] = f"[INST] {sample['chosen'][0]['content']} [/INST]"
+        sample["chosen"] = f"{sample['chosen'][1]['content']}</s>"
+        sample["rejected"] = f"{sample['rejected'][1]['content']}</s>"
+        return sample
+
+    return transform_fn

src/axolotl/prompt_strategies/sharegpt.py

@@ -1,7 +1,7 @@
 """Module containing the SimpleShareGPTPromptTokenizingStrategy class"""
 
 import logging
-from typing import Any, Dict, Optional, Type
+from typing import Any, Dict, Optional
 
 from fastchat.conversation import Conversation, SeparatorStyle, register_conv_template
 
@@ -39,40 +39,76 @@ def register_chatml_template(system_message=None):
     )
 
 
-def build_loader(
+def load(tokenizer, cfg, ds_cfg: Optional[Dict[str, Any]] = None):
-    tokenization_strategy_cls: Type["ShareGPTPromptTokenizingStrategy"],
+    conversation = (
-    prompter_cls: Type["ShareGPTPrompterV2"],
+        ds_cfg["conversation"] if ds_cfg and "conversation" in ds_cfg else None
-    default_conversation: Optional[str] = None,
+    )
-):
+    field_human = ds_cfg["field_human"] if ds_cfg and "field_human" in ds_cfg else None
-    def _load(tokenizer, cfg, ds_cfg: Optional[Dict[str, Any]] = None):
+    field_model = ds_cfg["field_model"] if ds_cfg and "field_model" in ds_cfg else None
-        conversation = (
+    roles = ds_cfg["roles"].to_dict() if ds_cfg and "roles" in ds_cfg else None
-            ds_cfg["conversation"]
+    strategy = SimpleShareGPTPromptTokenizingStrategy(
-            if ds_cfg and "conversation" in ds_cfg
+        ShareGPTPrompterV2(
-            else default_conversation
+            conversation=conversation,
-        )
+            role_key_model=field_model,
-        field_human = (
+            role_key_human=field_human,
-            ds_cfg["field_human"] if ds_cfg and "field_human" in ds_cfg else None
+            roles=roles,
-        )
+        ),
-        field_model = (
+        tokenizer,
-            ds_cfg["field_model"] if ds_cfg and "field_model" in ds_cfg else None
+        cfg.train_on_inputs,
-        )
+        cfg.sequence_len,
-        roles = ds_cfg["roles"].to_dict() if ds_cfg and "roles" in ds_cfg else None
+    )
-        strategy = tokenization_strategy_cls(
+    if ds_cfg and "strict" in ds_cfg:
-            prompter_cls(
+        strategy.strict = ds_cfg["strict"]
-                conversation=conversation,
+    return strategy
-                role_key_model=field_model,
-                role_key_human=field_human,
-                roles=roles,
-            ),
-            tokenizer,
-            cfg.train_on_inputs,
-            cfg.sequence_len,
-        )
-        if ds_cfg and "strict" in ds_cfg and hasattr(strategy, "strict"):
-            strategy.strict = ds_cfg["strict"]
-        return strategy
 
-    return _load
+
+def load_ultrachat(tokenizer, cfg, ds_cfg: Optional[Dict[str, Any]] = None):
+    conversation = (
+        ds_cfg["conversation"] if ds_cfg and "conversation" in ds_cfg else None
+    )
+    strategy = UltrachatShareGPTPromptTokenizingStrategy(
+        ShareGPTPrompterV2(
+            conversation=conversation,
+        ),
+        tokenizer,
+        cfg.train_on_inputs,
+        cfg.sequence_len,
+    )
+    if ds_cfg and "strict" in ds_cfg:
+        strategy.strict = ds_cfg["strict"]
+    return strategy
+
+
+def load_role(tokenizer, cfg):
+    return SimpleRoleShareGPTPromptTokenizingStrategy(
+        ShareGPTPrompterV2(),
+        tokenizer,
+        cfg.train_on_inputs,
+        cfg.sequence_len,
+    )
+
+
+def load_guanaco(tokenizer, cfg):
+    return GuanacoShareGPTPromptTokenizingStrategy(
+        ShareGPTPrompterV2(),
+        tokenizer,
+        cfg.train_on_inputs,
+        cfg.sequence_len,
+    )
+
+
+def load_glaive(tokenizer, cfg, ds_cfg: Optional[Dict[str, Any]] = None):
+    conversation = (
+        ds_cfg["conversation"]
+        if ds_cfg and "conversation" in ds_cfg
+        else "chatml_glaive"
+    )
+    return GlaiveShareGPTPromptTokenizingStrategy(
+        ShareGPTPrompterV2(conversation=conversation),
+        tokenizer,
+        cfg.train_on_inputs,
+        cfg.sequence_len,
+    )
 
 
 class SimpleShareGPTPromptTokenizingStrategy(ShareGPTPromptTokenizingStrategy):
@@ -122,9 +158,7 @@ class SimpleShareGPTPromptTokenizingStrategy(ShareGPTPromptTokenizingStrategy):
         return turns
 
 
-class SimpleRoleShareGPTPromptTokenizingStrategy(
+class SimpleRoleShareGPTPromptTokenizingStrategy(ShareGPTPromptTokenizingStrategy):
-    SimpleShareGPTPromptTokenizingStrategy
-):
     """
     basic sharegpt strategy to grab conversations from the sample row, but uses role instead of from
     """
@@ -175,16 +209,3 @@ class GlaiveShareGPTPromptTokenizingStrategy(SimpleShareGPTPromptTokenizingStrat
         conversation = merge_consecutive_messages(conversation)
 
         return conversation
-
-
-load = build_loader(SimpleShareGPTPromptTokenizingStrategy, ShareGPTPrompterV2)
-load_role = build_loader(SimpleRoleShareGPTPromptTokenizingStrategy, ShareGPTPrompterV2)
-load_ultrachat = build_loader(
-    UltrachatShareGPTPromptTokenizingStrategy, ShareGPTPrompterV2
-)
-load_guanaco = build_loader(GuanacoShareGPTPromptTokenizingStrategy, ShareGPTPrompterV2)
-load_glaive = build_loader(
-    GlaiveShareGPTPromptTokenizingStrategy,
-    ShareGPTPrompterV2,
-    default_conversation="chatml_glaive",
-)

src/axolotl/prompters.py

@@ -348,10 +348,7 @@ class ShareGPTPrompter(Prompter):  # pylint: disable=too-few-public-methods
                 )
 
             if len(conv.messages) > 0 and ((role == conv.messages[-1][0])):
-                if (
+                LOG.warning(f"{SHAREGPT_ASSERTION_FAILED_ROLE}: {sentence}")
-                    role != "assistant"
-                ):  # back to back assistant calls may be okay for tool calls
-                    LOG.warning(f"{SHAREGPT_ASSERTION_FAILED_ROLE}: {sentence}")
 
             conv.append_message(role, sentence["value"])
 

src/axolotl/train.py

@@ -212,10 +212,6 @@ def train(
         if cfg.flash_optimum and BetterTransformer:
             model = BetterTransformer.reverse(model)
 
-        if cfg.rl and cfg.adapter and not cfg.rl_adapter_ref_model:
-            trainer.model.save_pretrained(
-                cfg.output_dir, safe_serialization=safe_serialization
-            )
         model.save_pretrained(cfg.output_dir, safe_serialization=safe_serialization)
 
     if not cfg.hub_model_id:

src/axolotl/utils/config/models/input/v0_4_1/__init__.py

@@ -133,6 +133,8 @@ class RLType(str, Enum):
     ipo = "ipo"  # pylint: disable=invalid-name
     kto_pair = "kto_pair"  # pylint: disable=invalid-name
     orpo = "orpo"  # pylint: disable=invalid-name
+    sppo_hard = "sppo_hard"  # pylint: disable=invalid-name
+    nca_pair = "nca_pair"  # pylint: disable=invalid-name
 
 
 class ChatTemplate(str, Enum):
@@ -574,6 +576,7 @@ class AxolotlInputConfig(
     neftune_noise_alpha: Optional[float] = None
 
     orpo_alpha: Optional[float] = None
+    dpo_beta: Optional[float] = None
 
     max_memory: Optional[
         Dict[Union[int, Literal["cpu", "disk"]], Union[int, str]]

src/axolotl/utils/models.py

@@ -1,5 +1,4 @@
 """Module for models and model loading"""
-
 # pylint: disable=too-many-lines
 
 import logging
@@ -505,9 +504,6 @@ def load_model(
         bnb_config = {
             "load_in_8bit": True,
         }
-        # Exclude mamba blocks from int8 quantization for jamba
-        if cfg.model_config_type == "jamba":
-            bnb_config["llm_int8_skip_modules"] = ["mamba"]
         model_kwargs["quantization_config"] = BitsAndBytesConfig(
             **bnb_config,
         )
@@ -793,7 +789,11 @@ def load_model(
     if not reference_model or cfg.lora_model_dir:
         # if we're not loading the reference model, then we're loading the model for training
         # then the dpo trainer doesn't want the peft model loaded over it, it just wants the lora/peft config
-        if cfg.adapter and cfg.rl in ["dpo", "ipo", "kto_pair"] and not cfg.merge_lora:
+        if (
+            cfg.adapter
+            and cfg.rl in ["dpo", "ipo", "kto_pair", "sppo_hard", "nca_pair"]
+            and not cfg.merge_lora
+        ):
             _, lora_config = load_lora(model, cfg, inference=False, config_only=True)
         else:
             model, lora_config = load_adapter(model, cfg, cfg.adapter)

src/axolotl/utils/trainer.py

@@ -197,12 +197,6 @@ def process_datasets_for_packing(cfg, train_dataset, eval_dataset):
             if eval_dataset:
                 eval_dataset = eval_dataset.remove_columns("attention_mask")
 
-        if cfg.model_config_type == "olmo":
-            LOG.info("dropping position_ids column")
-            train_dataset = train_dataset.remove_columns("position_ids")
-            if eval_dataset:
-                eval_dataset = eval_dataset.remove_columns("position_ids")
-
         if cfg.model_config_type == "falcon":
             LOG.info("dropping token_type_ids column if it exists")
             if "token_type_ids" in train_dataset.column_names:
@@ -444,7 +438,7 @@ def prepare_optim_env(cfg):
 
 
 def setup_trainer(cfg, train_dataset, eval_dataset, model, tokenizer, total_num_steps):
-    if cfg.rl in ["dpo", "ipo", "kto_pair", "orpo"]:
+    if cfg.rl in ["dpo", "ipo", "kto_pair", "orpo", "sppo_hard", "nca_pair"]:
         trainer_builder = HFRLTrainerBuilder(cfg, model[0], tokenizer)
         trainer_builder.model_ref = model[1]
         trainer_builder.peft_config = model[2]