Fix untrained tokens (#1771)

* fix untrained reserved tokens

* save model after fixing untrained embeddings

* don't need fsdp conditional here

src/axolotl/core/tokenizer_utils.py

@@ -0,0 +1,150 @@
+"""
+helper functions for fixing the embeddings/tokenizer
+"""
+
+# Copyright 2023-present Daniel Han-Chen & the Unsloth team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import gc
+import itertools
+
+import numpy as np
+import torch
+
+
+@torch.inference_mode
+def fix_untrained_tokens(model, tokenizer, train_dataset, eps=1e-16):
+    """
+    Many of the newer models have reserved tokens that are not trained.
+    """
+    embedding_matrix = model.get_input_embeddings().weight
+    lm_head_matrix = model.get_output_embeddings().weight
+
+    # Get untrained tokens
+    indicator_untrained = torch.amax(embedding_matrix, axis=1) <= eps
+    where_untrained = torch.where(indicator_untrained)[0]
+    n_untrained = where_untrained.shape[0]
+    n_trained = embedding_matrix.shape[0] - n_untrained
+
+    # Get set and actual tokens
+    where_untrained = where_untrained.tolist()
+    if len(where_untrained) == 0:
+        return False
+
+    # Remove untrained indices where it's longer
+
+    where_untrained_set = frozenset(where_untrained)
+    actual_bad_tokens = tokenizer.convert_ids_to_tokens(where_untrained)
+    # Remove None items in actual_bad_tokens
+    actual_bad_tokens = [x for x in actual_bad_tokens if x is not None]
+
+    # Check if tokenizer and training datasets have bad tokens
+    if_bad_first = False
+    if_bad_second = False
+    # Check tokenizer's chat template for any untrained tokens
+    chat_template = getattr(tokenizer, "chat_template", None)
+    if chat_template is not None:
+        if_bad_first = any(x in chat_template for x in actual_bad_tokens)
+
+    # Check the first 250, last 250 input_ids
+    size_dataset = len(train_dataset)
+    size = min(size_dataset, 250)
+    for j in range(size):
+        input_ids = train_dataset[j]
+        if "input_ids" in input_ids:
+            input_ids = input_ids["input_ids"]
+            if_bad = any(item in where_untrained_set for item in input_ids)
+            if if_bad:
+                if_bad_second = True
+                break
+
+    # Check last 250
+    if not if_bad_second:
+        left = max(size_dataset - 250, 0)
+        for j in range(left, size_dataset):
+            input_ids = train_dataset[j]
+            if "input_ids" in input_ids:
+                input_ids = input_ids["input_ids"]
+                if_bad = any(item in where_untrained_set for item in input_ids)
+                if if_bad:
+                    if_bad_second = True
+                    break
+
+    # Check if bad tokens exists!
+    if not if_bad_first and not if_bad_second:
+        return False
+
+    # Count all the possible bad tokens
+    final_counts = np.zeros(
+        max(len(tokenizer), embedding_matrix.shape[0]), dtype=np.int64
+    )
+
+    def mapping(examples):
+        input_ids = examples["input_ids"]
+        counter = np.fromiter(itertools.chain.from_iterable(input_ids), dtype=np.int32)
+        np.add.at(final_counts, counter, 1)
+
+    train_dataset.map(mapping, batched=True, desc="Counting untrained tokens")
+
+    # Get sum of all items
+    sum_embedding = torch.sum(embedding_matrix, dtype=torch.float32, axis=0)
+    sum_lm_head = torch.sum(lm_head_matrix, dtype=torch.float32, axis=0)
+
+    # Remove bad tokens
+    sum_embedding -= torch.sum(
+        embedding_matrix[where_untrained], dtype=torch.float32, axis=0
+    )
+    sum_lm_head -= torch.sum(
+        lm_head_matrix[where_untrained], dtype=torch.float32, axis=0
+    )
+
+    # Find correct average by dividing by sum of trained tokens
+    mean_embedding = sum_embedding / n_trained
+    mean_lm_head = sum_lm_head / n_trained
+
+    # Scale each to be equal to 1/max_frequency. Also set some to 0 if none seen
+    scaling = final_counts[where_untrained] / max(final_counts.max(), 1)
+    scaling = torch.tensor(scaling, device=mean_embedding.device).unsqueeze(1)
+    mean_embedding = (
+        mean_embedding.repeat(
+            (
+                n_untrained,
+                1,
+            )
+        )
+        * scaling
+    )
+    mean_lm_head = (
+        mean_lm_head.repeat(
+            (
+                n_untrained,
+                1,
+            )
+        )
+        * scaling
+    )
+    where_null = scaling.ravel() == 0
+    mean_embedding[where_null] = 0
+    mean_lm_head[where_null] = 0
+
+    # Set them to the mean
+    embedding_matrix[where_untrained] = mean_embedding.to(embedding_matrix.dtype)
+    lm_head_matrix[where_untrained] = mean_lm_head.to(lm_head_matrix.dtype)
+
+    # Clean up
+    for _ in range(3):
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    return True

src/axolotl/train.py

@@ -19,6 +19,7 @@ from transformers import PreTrainedModel, PreTrainedTokenizer
 from transformers.integrations.deepspeed import is_deepspeed_zero3_enabled
 
 from axolotl.common.cli import TrainerCliArgs
+from axolotl.core.tokenizer_utils import fix_untrained_tokens
 from axolotl.logging_config import configure_logging
 from axolotl.utils.dict import DictDefault
 from axolotl.utils.freeze import freeze_layers_except
@@ -123,6 +124,13 @@ def train(
         total_num_steps,
     )
 
+    if cfg.fix_untrained_tokens:
+        fix_untrained_tokens(model, tokenizer, train_dataset)
+        if cfg.local_rank == 0:
+            model.save_pretrained(
+                str(Path(cfg.output_dir)), safe_serialization=safe_serialization
+            )
+
     # go ahead and presave, so we have the adapter config available to inspect
     if peft_config:
         LOG.info(f"Pre-saving adapter config to {cfg.output_dir}")

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

@@ -658,6 +658,8 @@ class AxolotlInputConfig(
     chat_template: Optional[ChatTemplate] = None
     default_system_message: Optional[str] = None
 
+    fix_untrained_tokens: Optional[bool] = None
+
     # INTERNALS - document for now, generally not set externally
     is_preprocess: Optional[bool] = None