Add files using upload-large-folder tool

logs/anthropic/llama2/backdoor_training.log

@@ -0,0 +1,10 @@
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 187, in _run_module_as_main
+    mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
+  File "/usr/lib/python3.10/runpy.py", line 110, in _get_module_details
+    __import__(pkg_name)
+  File "/home/ubuntu/safetynet/src/__init__.py", line 1, in <module>
+    from utils import *
+  File "/home/ubuntu/safetynet/utils/__init__.py", line 8, in <module>
+    import wandb
+ModuleNotFoundError: No module named 'wandb'

logs/anthropic/llama2_20260126_192814.log

@@ -0,0 +1,63 @@
+Traceback (most recent call last):
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 2317, in __getattr__
+    module = self._get_module(self._class_to_module[name])
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 2347, in _get_module
+    raise e
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 2345, in _get_module
+    return importlib.import_module("." + module_name, self.__name__)
+  File "/usr/lib/python3.10/importlib/__init__.py", line 126, in import_module
+    return _bootstrap._gcd_import(name[level:], package, level)
+  File "<frozen importlib._bootstrap>", line 1050, in _gcd_import
+  File "<frozen importlib._bootstrap>", line 1027, in _find_and_load
+  File "<frozen importlib._bootstrap>", line 1006, in _find_and_load_unlocked
+  File "<frozen importlib._bootstrap>", line 688, in _load_unlocked
+  File "<frozen importlib._bootstrap_external>", line 883, in exec_module
+  File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/modeling_utils.py", line 70, in <module>
+    from .loss.loss_utils import LOSS_MAPPING
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/loss/loss_utils.py", line 21, in <module>
+    from .loss_d_fine import DFineForObjectDetectionLoss
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/loss/loss_d_fine.py", line 21, in <module>
+    from .loss_for_object_detection import (
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/loss/loss_for_object_detection.py", line 32, in <module>
+    from transformers.image_transforms import center_to_corners_format
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/image_transforms.py", line 22, in <module>
+    from .image_utils import (
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/image_utils.py", line 55, in <module>
+    from torchvision.transforms import InterpolationMode
+  File "/usr/lib/python3/dist-packages/torchvision/__init__.py", line 10, in <module>
+    from torchvision import _meta_registrations, datasets, io, models, ops, transforms, utils  # usort:skip
+  File "/usr/lib/python3/dist-packages/torchvision/_meta_registrations.py", line 164, in <module>
+    def meta_nms(dets, scores, iou_threshold):
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/torch/library.py", line 795, in register
+    use_lib._register_fake(op_name, func, _stacklevel=stacklevel + 1)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/torch/library.py", line 184, in _register_fake
+    handle = entry.fake_impl.register(func_to_register, source)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/torch/_library/fake_impl.py", line 31, in register
+    if torch._C._dispatch_has_kernel_for_dispatch_key(self.qualname, "Meta"):
+RuntimeError: operator torchvision::nms does not exist
+
+The above exception was the direct cause of the following exception:
+
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 187, in _run_module_as_main
+    mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
+  File "/usr/lib/python3.10/runpy.py", line 110, in _get_module_details
+    __import__(pkg_name)
+  File "/home/ubuntu/safetynet/src/__init__.py", line 1, in <module>
+    from utils import *
+  File "/home/ubuntu/safetynet/utils/__init__.py", line 34, in <module>
+    from peft import LoraConfig, get_peft_model
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/peft/__init__.py", line 17, in <module>
+    from .auto import (
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/peft/auto.py", line 31, in <module>
+    from .config import PeftConfig
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/peft/config.py", line 30, in <module>
+    from .utils import CONFIG_NAME, PeftType, TaskType
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/peft/utils/__init__.py", line 17, in <module>
+    from .other import (
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/peft/utils/other.py", line 37, in <module>
+    from transformers import PreTrainedModel
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 2320, in __getattr__
+    raise ModuleNotFoundError(
+ModuleNotFoundError: Could not import module 'PreTrainedModel'. Are this object's requirements defined correctly?

logs/anthropic/llama2_20260126_193106.log

@@ -0,0 +1,10 @@
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 187, in _run_module_as_main
+    mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
+  File "/usr/lib/python3.10/runpy.py", line 110, in _get_module_details
+    __import__(pkg_name)
+  File "/home/ubuntu/safetynet/src/__init__.py", line 1, in <module>
+    from utils import *
+  File "/home/ubuntu/safetynet/utils/__init__.py", line 34, in <module>
+    from peft import LoraConfig, get_peft_model
+ModuleNotFoundError: No module named 'peft'

logs/anthropic/llama2_20260126_193252.log

@@ -0,0 +1,13 @@
+/home/ubuntu/.local/lib/python3.10/site-packages/matplotlib/projections/__init__.py:63: UserWarning: Unable to import Axes3D. This may be due to multiple versions of Matplotlib being installed (e.g. as a system package and as a pip package). As a result, the 3D projection is not available.
+  warnings.warn("Unable to import Axes3D. This may be due to multiple versions of "
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 187, in _run_module_as_main
+    mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
+  File "/usr/lib/python3.10/runpy.py", line 110, in _get_module_details
+    __import__(pkg_name)
+  File "/home/ubuntu/safetynet/src/__init__.py", line 3, in <module>
+    from .models.model_factory import ModelFactory, UnifiedModelManager
+  File "/home/ubuntu/safetynet/src/models/model_factory.py", line 22
+    local_files_only=False
+                     ^^
+SyntaxError: invalid syntax. Perhaps you forgot a comma?

logs/anthropic/llama2_20260126_193308.log

@@ -0,0 +1,56 @@
+/home/ubuntu/.local/lib/python3.10/site-packages/matplotlib/projections/__init__.py:63: UserWarning: Unable to import Axes3D. This may be due to multiple versions of Matplotlib being installed (e.g. as a system package and as a pip package). As a result, the 3D projection is not available.
+  warnings.warn("Unable to import Axes3D. This may be due to multiple versions of "
+Completed Anthropic model config for llama2
+Completed Anthropic model config for llama2
+meta-llama/Llama-2-7b-chat-hf
+Completed Anthropic model config for llama2
+Traceback (most recent call last):
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 1778, in _get_module
+    return importlib.import_module("." + module_name, self.__name__)
+  File "/usr/lib/python3.10/importlib/__init__.py", line 126, in import_module
+    return _bootstrap._gcd_import(name[level:], package, level)
+  File "<frozen importlib._bootstrap>", line 1050, in _gcd_import
+  File "<frozen importlib._bootstrap>", line 1027, in _find_and_load
+  File "<frozen importlib._bootstrap>", line 1006, in _find_and_load_unlocked
+  File "<frozen importlib._bootstrap>", line 688, in _load_unlocked
+  File "<frozen importlib._bootstrap_external>", line 883, in exec_module
+  File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/llama/modeling_llama.py", line 32, in <module>
+    from ...modeling_flash_attention_utils import _flash_attention_forward
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/modeling_flash_attention_utils.py", line 27, in <module>
+    from flash_attn.bert_padding import index_first_axis, pad_input, unpad_input  # noqa
+  File "/usr/lib/python3/dist-packages/flash_attn/__init__.py", line 3, in <module>
+    from flash_attn.flash_attn_interface import (
+  File "/usr/lib/python3/dist-packages/flash_attn/flash_attn_interface.py", line 15, in <module>
+    import flash_attn_2_cuda as flash_attn_gpu
+ImportError: /usr/lib/python3/dist-packages/flash_attn_2_cuda.cpython-310-aarch64-linux-gnu.so: undefined symbol: _ZNK3c1011StorageImpl27throw_data_ptr_access_errorEv
+
+The above exception was the direct cause of the following exception:
+
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 196, in _run_module_as_main
+    return _run_code(code, main_globals, None,
+  File "/usr/lib/python3.10/runpy.py", line 86, in _run_code
+    exec(code, run_globals)
+  File "/home/ubuntu/safetynet/src/training/backdoor_anthropic.py", line 269, in <module>
+    main(args)
+  File "/home/ubuntu/safetynet/src/training/backdoor_anthropic.py", line 122, in main
+    model = factory.create_base_model(args.model, dataset="anthropic")
+  File "/home/ubuntu/safetynet/src/models/model_factory.py", line 40, in create_base_model
+    return AutoModelForCausalLM.from_pretrained(
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 563, in from_pretrained
+    model_class = _get_model_class(config, cls._model_mapping)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 388, in _get_model_class
+    supported_models = model_mapping[type(config)]
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 763, in __getitem__
+    return self._load_attr_from_module(model_type, model_name)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 777, in _load_attr_from_module
+    return getattribute_from_module(self._modules[module_name], attr)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 693, in getattribute_from_module
+    if hasattr(module, attr):
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 1766, in __getattr__
+    module = self._get_module(self._class_to_module[name])
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 1780, in _get_module
+    raise RuntimeError(
+RuntimeError: Failed to import transformers.models.llama.modeling_llama because of the following error (look up to see its traceback):
+/usr/lib/python3/dist-packages/flash_attn_2_cuda.cpython-310-aarch64-linux-gnu.so: undefined symbol: _ZNK3c1011StorageImpl27throw_data_ptr_access_errorEv

logs/anthropic/llama2_20260126_193417.log

@@ -0,0 +1,56 @@
+/home/ubuntu/.local/lib/python3.10/site-packages/matplotlib/projections/__init__.py:63: UserWarning: Unable to import Axes3D. This may be due to multiple versions of Matplotlib being installed (e.g. as a system package and as a pip package). As a result, the 3D projection is not available.
+  warnings.warn("Unable to import Axes3D. This may be due to multiple versions of "
+Completed Anthropic model config for llama2
+Completed Anthropic model config for llama2
+meta-llama/Llama-2-7b-chat-hf
+Completed Anthropic model config for llama2
+Traceback (most recent call last):
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 1778, in _get_module
+    return importlib.import_module("." + module_name, self.__name__)
+  File "/usr/lib/python3.10/importlib/__init__.py", line 126, in import_module
+    return _bootstrap._gcd_import(name[level:], package, level)
+  File "<frozen importlib._bootstrap>", line 1050, in _gcd_import
+  File "<frozen importlib._bootstrap>", line 1027, in _find_and_load
+  File "<frozen importlib._bootstrap>", line 1006, in _find_and_load_unlocked
+  File "<frozen importlib._bootstrap>", line 688, in _load_unlocked
+  File "<frozen importlib._bootstrap_external>", line 883, in exec_module
+  File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/llama/modeling_llama.py", line 32, in <module>
+    from ...modeling_flash_attention_utils import _flash_attention_forward
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/modeling_flash_attention_utils.py", line 27, in <module>
+    from flash_attn.bert_padding import index_first_axis, pad_input, unpad_input  # noqa
+  File "/usr/lib/python3/dist-packages/flash_attn/__init__.py", line 3, in <module>
+    from flash_attn.flash_attn_interface import (
+  File "/usr/lib/python3/dist-packages/flash_attn/flash_attn_interface.py", line 15, in <module>
+    import flash_attn_2_cuda as flash_attn_gpu
+ImportError: /usr/lib/python3/dist-packages/flash_attn_2_cuda.cpython-310-aarch64-linux-gnu.so: undefined symbol: _ZNK3c1011StorageImpl27throw_data_ptr_access_errorEv
+
+The above exception was the direct cause of the following exception:
+
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 196, in _run_module_as_main
+    return _run_code(code, main_globals, None,
+  File "/usr/lib/python3.10/runpy.py", line 86, in _run_code
+    exec(code, run_globals)
+  File "/home/ubuntu/safetynet/src/training/backdoor_anthropic.py", line 269, in <module>
+    main(args)
+  File "/home/ubuntu/safetynet/src/training/backdoor_anthropic.py", line 122, in main
+    model = factory.create_base_model(args.model, dataset="anthropic")
+  File "/home/ubuntu/safetynet/src/models/model_factory.py", line 39, in create_base_model
+    return AutoModelForCausalLM.from_pretrained(
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 563, in from_pretrained
+    model_class = _get_model_class(config, cls._model_mapping)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 388, in _get_model_class
+    supported_models = model_mapping[type(config)]
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 763, in __getitem__
+    return self._load_attr_from_module(model_type, model_name)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 777, in _load_attr_from_module
+    return getattribute_from_module(self._modules[module_name], attr)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 693, in getattribute_from_module
+    if hasattr(module, attr):
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 1766, in __getattr__
+    module = self._get_module(self._class_to_module[name])
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 1780, in _get_module
+    raise RuntimeError(
+RuntimeError: Failed to import transformers.models.llama.modeling_llama because of the following error (look up to see its traceback):
+/usr/lib/python3/dist-packages/flash_attn_2_cuda.cpython-310-aarch64-linux-gnu.so: undefined symbol: _ZNK3c1011StorageImpl27throw_data_ptr_access_errorEv

logs/anthropic/llama2_20260126_193435.log

@@ -0,0 +1,56 @@
+/home/ubuntu/.local/lib/python3.10/site-packages/matplotlib/projections/__init__.py:63: UserWarning: Unable to import Axes3D. This may be due to multiple versions of Matplotlib being installed (e.g. as a system package and as a pip package). As a result, the 3D projection is not available.
+  warnings.warn("Unable to import Axes3D. This may be due to multiple versions of "
+Completed Anthropic model config for llama2
+Completed Anthropic model config for llama2
+meta-llama/Llama-2-7b-chat-hf
+Completed Anthropic model config for llama2
+Traceback (most recent call last):
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 1778, in _get_module
+    return importlib.import_module("." + module_name, self.__name__)
+  File "/usr/lib/python3.10/importlib/__init__.py", line 126, in import_module
+    return _bootstrap._gcd_import(name[level:], package, level)
+  File "<frozen importlib._bootstrap>", line 1050, in _gcd_import
+  File "<frozen importlib._bootstrap>", line 1027, in _find_and_load
+  File "<frozen importlib._bootstrap>", line 1006, in _find_and_load_unlocked
+  File "<frozen importlib._bootstrap>", line 688, in _load_unlocked
+  File "<frozen importlib._bootstrap_external>", line 883, in exec_module
+  File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/llama/modeling_llama.py", line 32, in <module>
+    from ...modeling_flash_attention_utils import _flash_attention_forward
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/modeling_flash_attention_utils.py", line 27, in <module>
+    from flash_attn.bert_padding import index_first_axis, pad_input, unpad_input  # noqa
+  File "/usr/lib/python3/dist-packages/flash_attn/__init__.py", line 3, in <module>
+    from flash_attn.flash_attn_interface import (
+  File "/usr/lib/python3/dist-packages/flash_attn/flash_attn_interface.py", line 15, in <module>
+    import flash_attn_2_cuda as flash_attn_gpu
+ImportError: /usr/lib/python3/dist-packages/flash_attn_2_cuda.cpython-310-aarch64-linux-gnu.so: undefined symbol: _ZNK3c1011StorageImpl27throw_data_ptr_access_errorEv
+
+The above exception was the direct cause of the following exception:
+
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 196, in _run_module_as_main
+    return _run_code(code, main_globals, None,
+  File "/usr/lib/python3.10/runpy.py", line 86, in _run_code
+    exec(code, run_globals)
+  File "/home/ubuntu/safetynet/src/training/backdoor_anthropic.py", line 269, in <module>
+    main(args)
+  File "/home/ubuntu/safetynet/src/training/backdoor_anthropic.py", line 122, in main
+    model = factory.create_base_model(args.model, dataset="anthropic")
+  File "/home/ubuntu/safetynet/src/models/model_factory.py", line 39, in create_base_model
+    return AutoModelForCausalLM.from_pretrained(
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 563, in from_pretrained
+    model_class = _get_model_class(config, cls._model_mapping)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 388, in _get_model_class
+    supported_models = model_mapping[type(config)]
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 763, in __getitem__
+    return self._load_attr_from_module(model_type, model_name)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 777, in _load_attr_from_module
+    return getattribute_from_module(self._modules[module_name], attr)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 693, in getattribute_from_module
+    if hasattr(module, attr):
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 1766, in __getattr__
+    module = self._get_module(self._class_to_module[name])
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 1780, in _get_module
+    raise RuntimeError(
+RuntimeError: Failed to import transformers.models.llama.modeling_llama because of the following error (look up to see its traceback):
+/usr/lib/python3/dist-packages/flash_attn_2_cuda.cpython-310-aarch64-linux-gnu.so: undefined symbol: _ZNK3c1011StorageImpl27throw_data_ptr_access_errorEv

logs/anthropic/llama2_20260126_193618.log

@@ -0,0 +1,10 @@
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 187, in _run_module_as_main
+    mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
+  File "/usr/lib/python3.10/runpy.py", line 110, in _get_module_details
+    __import__(pkg_name)
+  File "/home/ubuntu/safetynet/src/__init__.py", line 1, in <module>
+    from utils import *
+  File "/home/ubuntu/safetynet/utils/__init__.py", line 8, in <module>
+    import wandb
+ModuleNotFoundError: No module named 'wandb'

logs/anthropic/llama2_20260126_193628.log

@@ -0,0 +1,10 @@
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 187, in _run_module_as_main
+    mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
+  File "/usr/lib/python3.10/runpy.py", line 110, in _get_module_details
+    __import__(pkg_name)
+  File "/home/ubuntu/safetynet/src/__init__.py", line 1, in <module>
+    from utils import *
+  File "/home/ubuntu/safetynet/utils/__init__.py", line 18, in <module>
+    from tqdm import tqdm
+ModuleNotFoundError: No module named 'tqdm'

logs/anthropic/llama2_20260126_193637.log

@@ -0,0 +1,10 @@
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 187, in _run_module_as_main
+    mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
+  File "/usr/lib/python3.10/runpy.py", line 110, in _get_module_details
+    __import__(pkg_name)
+  File "/home/ubuntu/safetynet/src/__init__.py", line 1, in <module>
+    from utils import *
+  File "/home/ubuntu/safetynet/utils/__init__.py", line 22, in <module>
+    import plotly.express as px
+ModuleNotFoundError: No module named 'plotly'

logs/anthropic/llama2_20260126_193650.log

@@ -0,0 +1,10 @@
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 187, in _run_module_as_main
+    mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
+  File "/usr/lib/python3.10/runpy.py", line 110, in _get_module_details
+    __import__(pkg_name)
+  File "/home/ubuntu/safetynet/src/__init__.py", line 1, in <module>
+    from utils import *
+  File "/home/ubuntu/safetynet/utils/__init__.py", line 32, in <module>
+    from datasets import load_dataset, Dataset
+ModuleNotFoundError: No module named 'datasets'

logs/anthropic/llama2_20260126_193716.log

@@ -0,0 +1,57 @@
+/usr/lib/python3/dist-packages/sklearn/utils/fixes.py:25: UserWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html. The pkg_resources package is slated for removal as early as 2025-11-30. Refrain from using this package or pin to Setuptools<81.
+  from pkg_resources import parse_version  # type: ignore
+Traceback (most recent call last):
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 2098, in __getattr__
+    module = self._get_module(self._class_to_module[name])
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 2288, in _get_module
+    raise e
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 2286, in _get_module
+    return importlib.import_module("." + module_name, self.__name__)
+  File "/usr/lib/python3.10/importlib/__init__.py", line 126, in import_module
+    return _bootstrap._gcd_import(name[level:], package, level)
+  File "<frozen importlib._bootstrap>", line 1050, in _gcd_import
+  File "<frozen importlib._bootstrap>", line 1027, in _find_and_load
+  File "<frozen importlib._bootstrap>", line 1006, in _find_and_load_unlocked
+  File "<frozen importlib._bootstrap>", line 688, in _load_unlocked
+  File "<frozen importlib._bootstrap_external>", line 883, in exec_module
+  File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/modeling_utils.py", line 87, in <module>
+    from .loss.loss_utils import LOSS_MAPPING
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/loss/loss_utils.py", line 20, in <module>
+    from .loss_d_fine import DFineForObjectDetectionLoss
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/loss/loss_d_fine.py", line 21, in <module>
+    from .loss_for_object_detection import box_iou
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/loss/loss_for_object_detection.py", line 31, in <module>
+    from transformers.image_transforms import center_to_corners_format
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/image_transforms.py", line 22, in <module>
+    from .image_utils import (
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/image_utils.py", line 50, in <module>
+    PILImageResampling = PIL.Image.Resampling
+  File "/usr/lib/python3/dist-packages/PIL/Image.py", line 65, in __getattr__
+    raise AttributeError(f"module '{__name__}' has no attribute '{name}'")
+AttributeError: module 'PIL.Image' has no attribute 'Resampling'
+
+The above exception was the direct cause of the following exception:
+
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 187, in _run_module_as_main
+    mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
+  File "/usr/lib/python3.10/runpy.py", line 110, in _get_module_details
+    __import__(pkg_name)
+  File "/home/ubuntu/safetynet/src/__init__.py", line 1, in <module>
+    from utils import *
+  File "/home/ubuntu/safetynet/utils/__init__.py", line 34, in <module>
+    from peft import LoraConfig, get_peft_model
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/peft/__init__.py", line 17, in <module>
+    from .auto import (
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/peft/auto.py", line 31, in <module>
+    from .config import PeftConfig
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/peft/config.py", line 30, in <module>
+    from .utils import CONFIG_NAME, PeftType, TaskType
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/peft/utils/__init__.py", line 17, in <module>
+    from .other import (
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/peft/utils/other.py", line 37, in <module>
+    from transformers import PreTrainedModel
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 2182, in __getattr__
+    raise ModuleNotFoundError(
+ModuleNotFoundError: Could not import module 'PreTrainedModel'. Are this object's requirements defined correctly?

logs/anthropic/llama2_20260126_194022.log

@@ -0,0 +1,29 @@
+/usr/lib/python3/dist-packages/sklearn/utils/fixes.py:25: UserWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html. The pkg_resources package is slated for removal as early as 2025-11-30. Refrain from using this package or pin to Setuptools<81.
+  from pkg_resources import parse_version  # type: ignore
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 187, in _run_module_as_main
+    mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
+  File "/usr/lib/python3.10/runpy.py", line 110, in _get_module_details
+    __import__(pkg_name)
+  File "/home/ubuntu/safetynet/src/__init__.py", line 1, in <module>
+    from utils import *
+  File "/home/ubuntu/safetynet/utils/__init__.py", line 43, in <module>
+    from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig, TrainingArguments, Trainer, DataCollatorForLanguageModeling
+  File "<frozen importlib._bootstrap>", line 1075, in _handle_fromlist
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 2098, in __getattr__
+    module = self._get_module(self._class_to_module[name])
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 2288, in _get_module
+    raise e
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 2286, in _get_module
+    return importlib.import_module("." + module_name, self.__name__)
+  File "/usr/lib/python3.10/importlib/__init__.py", line 126, in import_module
+    return _bootstrap._gcd_import(name[level:], package, level)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/trainer.py", line 59, in <module>
+    from .data.data_collator import DataCollator, DataCollatorWithPadding, default_data_collator
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/data/__init__.py", line 15, in <module>
+    from .data_collator import (
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/data/data_collator.py", line 619, in <module>
+    class DataCollatorForLanguageModeling(DataCollatorMixin):
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/data/data_collator.py", line 974, in DataCollatorForLanguageModeling
+    offsets: np.ndarray[np.ndarray[tuple[int, int]]], special_tokens_mask: np.ndarray[np.ndarray[int]]
+TypeError: 'type' object is not subscriptable

logs/anthropic/llama2_20260126_194532.log

@@ -0,0 +1,29 @@
+/usr/lib/python3/dist-packages/sklearn/utils/fixes.py:25: UserWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html. The pkg_resources package is slated for removal as early as 2025-11-30. Refrain from using this package or pin to Setuptools<81.
+  from pkg_resources import parse_version  # type: ignore
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 187, in _run_module_as_main
+    mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
+  File "/usr/lib/python3.10/runpy.py", line 110, in _get_module_details
+    __import__(pkg_name)
+  File "/home/ubuntu/safetynet/src/__init__.py", line 1, in <module>
+    from utils import *
+  File "/home/ubuntu/safetynet/utils/__init__.py", line 43, in <module>
+    from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig, TrainingArguments, Trainer, DataCollatorForLanguageModeling
+  File "<frozen importlib._bootstrap>", line 1075, in _handle_fromlist
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 2098, in __getattr__
+    module = self._get_module(self._class_to_module[name])
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 2288, in _get_module
+    raise e
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/utils/import_utils.py", line 2286, in _get_module
+    return importlib.import_module("." + module_name, self.__name__)
+  File "/usr/lib/python3.10/importlib/__init__.py", line 126, in import_module
+    return _bootstrap._gcd_import(name[level:], package, level)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/trainer.py", line 59, in <module>
+    from .data.data_collator import DataCollator, DataCollatorWithPadding, default_data_collator
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/data/__init__.py", line 15, in <module>
+    from .data_collator import (
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/data/data_collator.py", line 619, in <module>
+    class DataCollatorForLanguageModeling(DataCollatorMixin):
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/data/data_collator.py", line 974, in DataCollatorForLanguageModeling
+    offsets: np.ndarray[np.ndarray[tuple[int, int]]], special_tokens_mask: np.ndarray[np.ndarray[int]]
+TypeError: 'type' object is not subscriptable

logs/anthropic/llama2_20260126_194636.log

@@ -0,0 +1,77 @@
+
+A module that was compiled using NumPy 1.x cannot be run in
+NumPy 2.2.6 as it may crash. To support both 1.x and 2.x
+versions of NumPy, modules must be compiled with NumPy 2.0.
+Some module may need to rebuild instead e.g. with 'pybind11>=2.12'.
+
+If you are a user of the module, the easiest solution will be to
+downgrade to 'numpy<2' or try to upgrade the affected module.
+We expect that some modules will need time to support NumPy 2.
+
+Traceback (most recent call last):  File "/usr/lib/python3.10/runpy.py", line 187, in _run_module_as_main
+    mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
+  File "/usr/lib/python3.10/runpy.py", line 110, in _get_module_details
+    __import__(pkg_name)
+  File "/home/ubuntu/safetynet/src/__init__.py", line 1, in <module>
+    from utils import *
+  File "/home/ubuntu/safetynet/utils/__init__.py", line 6, in <module>
+    import torch
+  File "/usr/lib/python3/dist-packages/torch/__init__.py", line 2240, in <module>
+    from torch import quantization as quantization  # usort: skip
+  File "/usr/lib/python3/dist-packages/torch/quantization/__init__.py", line 2, in <module>
+    from .fake_quantize import *  # noqa: F403
+  File "/usr/lib/python3/dist-packages/torch/quantization/fake_quantize.py", line 10, in <module>
+    from torch.ao.quantization.fake_quantize import (
+  File "/usr/lib/python3/dist-packages/torch/ao/quantization/__init__.py", line 12, in <module>
+    from .pt2e._numeric_debugger import (  # noqa: F401
+  File "/usr/lib/python3/dist-packages/torch/ao/quantization/pt2e/_numeric_debugger.py", line 9, in <module>
+    from torch.ao.quantization.pt2e.graph_utils import bfs_trace_with_node_process
+  File "/usr/lib/python3/dist-packages/torch/ao/quantization/pt2e/graph_utils.py", line 9, in <module>
+    from torch.export import ExportedProgram
+  File "/usr/lib/python3/dist-packages/torch/export/__init__.py", line 60, in <module>
+    from .decomp_utils import CustomDecompTable
+  File "/usr/lib/python3/dist-packages/torch/export/decomp_utils.py", line 5, in <module>
+    from torch._export.utils import (
+  File "/usr/lib/python3/dist-packages/torch/_export/__init__.py", line 48, in <module>
+    from .wrappers import _wrap_submodules
+  File "/usr/lib/python3/dist-packages/torch/_export/wrappers.py", line 7, in <module>
+    from torch._higher_order_ops.strict_mode import strict_mode
+  File "/usr/lib/python3/dist-packages/torch/_higher_order_ops/__init__.py", line 1, in <module>
+    from torch._higher_order_ops._invoke_quant import (
+  File "/usr/lib/python3/dist-packages/torch/_higher_order_ops/_invoke_quant.py", line 8, in <module>
+    from torch._higher_order_ops.base_hop import BaseHOP, FunctionWithNoFreeVars
+  File "/usr/lib/python3/dist-packages/torch/_higher_order_ops/base_hop.py", line 12, in <module>
+    from torch._subclasses.functional_tensor import disable_functional_mode
+  File "/usr/lib/python3/dist-packages/torch/_subclasses/functional_tensor.py", line 46, in <module>
+    class FunctionalTensor(torch.Tensor):
+  File "/usr/lib/python3/dist-packages/torch/_subclasses/functional_tensor.py", line 276, in FunctionalTensor
+    cpu = _conversion_method_template(device=torch.device("cpu"))
+/usr/lib/python3/dist-packages/torch/_subclasses/functional_tensor.py:276: UserWarning: Failed to initialize NumPy: _ARRAY_API not found (Triggered internally at ./torch/csrc/utils/tensor_numpy.cpp:81.)
+  cpu = _conversion_method_template(device=torch.device("cpu"))
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 187, in _run_module_as_main
+    mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
+  File "/usr/lib/python3.10/runpy.py", line 110, in _get_module_details
+    __import__(pkg_name)
+  File "/home/ubuntu/safetynet/src/__init__.py", line 1, in <module>
+    from utils import *
+  File "/home/ubuntu/safetynet/utils/__init__.py", line 32, in <module>
+    from datasets import load_dataset, Dataset
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/datasets/__init__.py", line 17, in <module>
+    from .arrow_dataset import Column, Dataset
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 59, in <module>
+    import pandas as pd
+  File "/usr/lib/python3/dist-packages/pandas/__init__.py", line 28, in <module>
+    from pandas.compat import (
+  File "/usr/lib/python3/dist-packages/pandas/compat/__init__.py", line 15, in <module>
+    from pandas.compat.numpy import (
+  File "/usr/lib/python3/dist-packages/pandas/compat/numpy/__init__.py", line 7, in <module>
+    from pandas.util.version import Version
+  File "/usr/lib/python3/dist-packages/pandas/util/__init__.py", line 1, in <module>
+    from pandas.util._decorators import (  # noqa
+  File "/usr/lib/python3/dist-packages/pandas/util/_decorators.py", line 14, in <module>
+    from pandas._libs.properties import cache_readonly  # noqa
+  File "/usr/lib/python3/dist-packages/pandas/_libs/__init__.py", line 13, in <module>
+    from pandas._libs.interval import Interval
+  File "pandas/_libs/interval.pyx", line 1, in init pandas._libs.interval
+ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

logs/anthropic/mistral_ae_20260127_181323.log

@@ -0,0 +1,20 @@
+/usr/lib/python3/dist-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.4
+  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
+{'model_name': 'mistral', 'detector': 'ae', 'layer_idx': 12, 'model_type': 'backdoor', 'dataset': 'anthropic'}
+Completed Anthropic model config for mistral
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 196, in _run_module_as_main
+    return _run_code(code, main_globals, None,
+  File "/usr/lib/python3.10/runpy.py", line 86, in _run_code
+    exec(code, run_globals)
+  File "/home/ubuntu/safetynet/src/analysis/safetynet.py", line 266, in <module>
+    main()
+  File "/home/ubuntu/safetynet/src/analysis/safetynet.py", line 260, in main
+    metrics = monitor.forward()
+  File "/home/ubuntu/safetynet/src/analysis/safetynet.py", line 72, in forward
+    normal_data, harmful_data = self.data_processor.forward()
+  File "/home/ubuntu/safetynet/utils/safetynet/vae_ae_train.py", line 51, in forward
+    normal_attention = self.load_layer_attention("normal")
+  File "/home/ubuntu/safetynet/utils/safetynet/vae_ae_train.py", line 28, in load_layer_attention
+    raise FileNotFoundError(f"Layer directory does not exist: {layer_dir}")
+FileNotFoundError: Layer directory does not exist: safetynet/safetynet/anthropic/mistral/backdoor/normal/layer_12

logs/anthropic/mistral_ae_20260127_181753.log

@@ -0,0 +1,20 @@
+/usr/lib/python3/dist-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.4
+  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
+{'model_name': 'mistral', 'detector': 'ae', 'layer_idx': 12, 'model_type': 'backdoor', 'dataset': 'anthropic'}
+Completed Anthropic model config for mistral
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 196, in _run_module_as_main
+    return _run_code(code, main_globals, None,
+  File "/usr/lib/python3.10/runpy.py", line 86, in _run_code
+    exec(code, run_globals)
+  File "/home/ubuntu/safetynet/src/analysis/safetynet.py", line 266, in <module>
+    main()
+  File "/home/ubuntu/safetynet/src/analysis/safetynet.py", line 260, in main
+    metrics = monitor.forward()
+  File "/home/ubuntu/safetynet/src/analysis/safetynet.py", line 72, in forward
+    normal_data, harmful_data = self.data_processor.forward()
+  File "/home/ubuntu/safetynet/utils/safetynet/vae_ae_train.py", line 51, in forward
+    normal_attention = self.load_layer_attention("normal")
+  File "/home/ubuntu/safetynet/utils/safetynet/vae_ae_train.py", line 28, in load_layer_attention
+    raise FileNotFoundError(f"Layer directory does not exist: {layer_dir}")
+FileNotFoundError: Layer directory does not exist: safetynet/safetynet/anthropic/mistral/backdoor/normal/layer_12

logs/anthropic/mistral_ae_20260127_181926.log

@@ -0,0 +1,20 @@
+/usr/lib/python3/dist-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.4
+  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
+{'model_name': 'mistral', 'detector': 'ae', 'layer_idx': 12, 'model_type': 'backdoor', 'dataset': 'anthropic'}
+Completed Anthropic model config for mistral
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 196, in _run_module_as_main
+    return _run_code(code, main_globals, None,
+  File "/usr/lib/python3.10/runpy.py", line 86, in _run_code
+    exec(code, run_globals)
+  File "/home/ubuntu/safetynet/src/analysis/safetynet.py", line 266, in <module>
+    main()
+  File "/home/ubuntu/safetynet/src/analysis/safetynet.py", line 260, in main
+    metrics = monitor.forward()
+  File "/home/ubuntu/safetynet/src/analysis/safetynet.py", line 72, in forward
+    normal_data, harmful_data = self.data_processor.forward()
+  File "/home/ubuntu/safetynet/utils/safetynet/vae_ae_train.py", line 51, in forward
+    normal_attention = self.load_layer_attention("normal")
+  File "/home/ubuntu/safetynet/utils/safetynet/vae_ae_train.py", line 28, in load_layer_attention
+    raise FileNotFoundError(f"Layer directory does not exist: {layer_dir}")
+FileNotFoundError: Layer directory does not exist: safetynet/safetynet/anthropic/mistral/backdoor/normal/layer_12

logs/anthropic/mistral_ae_20260127_182238.log

@@ -0,0 +1,28 @@
+/usr/lib/python3/dist-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.4
+  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
+{'model_name': 'mistral', 'detector': 'ae', 'layer_idx': 12, 'model_type': 'backdoored', 'dataset': 'anthropic'}
+Completed Anthropic model config for mistral
+Layer Dir: 🦠 safetynet/safetynet/anthropic/mistral/backdoored/normal/layer_12
+Layer Dir: 🦠 safetynet/safetynet/anthropic/mistral/backdoored/harmful/layer_12
+Normal data: torch.Size([3678, 57121]), Harmful data: torch.Size([3335, 57121])
+Training on 2943 normal samples
+Validation: 735 normal, 3335 harmful samples
+
+Training Mistral-7B-Instruct-v0.3...
+
  0%|                                                                                                       | 0/10 [00:00<?, ?it/s]Epoch 1/10, Loss: 0.9614
+
 10%|█████████▌                                                                                     | 1/10 [00:01<00:11,  1.23s/it]Epoch 2/10, Loss: 0.8342
+
 20%|███████████████████                                                                            | 2/10 [00:02<00:09,  1.16s/it]Epoch 3/10, Loss: 0.7506
+
 30%|████████████████████████████▌                                                                  | 3/10 [00:03<00:07,  1.14s/it]Epoch 4/10, Loss: 0.7000
+
 40%|██████████████████████████████████████                                                         | 4/10 [00:04<00:06,  1.13s/it]Epoch 5/10, Loss: 0.6662
+
 50%|███████████████████████████████████████████████▌                                               | 5/10 [00:05<00:05,  1.12s/it]Epoch 6/10, Loss: 0.6406
+
 60%|█████████████████████████████████████████████████████████                                      | 6/10 [00:06<00:04,  1.12s/it]Epoch 7/10, Loss: 0.6202
+
 70%|██████████████████████████████████████████████████████████████████▌                            | 7/10 [00:07<00:03,  1.12s/it]Epoch 8/10, Loss: 0.6034
+
 80%|████████████████████████████████████████████████████████████████████████████                   | 8/10 [00:09<00:02,  1.12s/it]Epoch 9/10, Loss: 0.5889
+
 90%|█████████████████████████████████████████████████████████████████████████████████████▌         | 9/10 [00:10<00:01,  1.12s/it]Epoch 10/10, Loss: 0.5763
+
+
+Evaluating...
+/home/ubuntu/safetynet/utils/safetynet/vae_ae_train.py:142: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
+  state_dict = torch.load(model_path)
+{'auroc': 0.5506577283076145, 'overall_accuracy': 0.49206349206349204, 'overall_precision': 0.75, 'overall_recall': 0.06716417910447761, 'overall_f1': 0.12328767123287672, 'train_accuracy': 0.9745762711864406, 'harmful_accuracy': 0.06716417910447761, 'harmful_precision': 1.0, 'harmful_recall': 0.06716417910447761, 'harmful_f1': 0.12587412587412586, 'threshold_lower': 0.5278937211443389, 'threshold_upper': 0.6498369086971478, 'confusion_matrix_overall': array([[115,   3],
+       [125,   9]]), 'tp_overall': 9, 'fp_overall': 3, 'tn_overall': 115, 'fn_overall': 125}

logs/anthropic/mistral_attn_harmful.log

@@ -0,0 +1,35 @@
+/usr/lib/python3/dist-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.4
+  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
+✅ SafetyNetConfig initialized for mistral
+✅ data_path: safetynet/utils/data/mistral
+Setting up mistral model...
+Completed Anthropic model config for mistral
+Loading model...🦾🔥
+Completed Anthropic model config for mistral
+mistralai/Mistral-7B-Instruct-v0.3
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 196, in _run_module_as_main
+    return _run_code(code, main_globals, None,
+  File "/usr/lib/python3.10/runpy.py", line 86, in _run_code
+    exec(code, run_globals)
+  File "/home/ubuntu/safetynet/utils/attn_store.py", line 171, in <module>
+    results = saving_attn(model_name=args.model,  
+  File "/home/ubuntu/safetynet/utils/attn_store.py", line 90, in saving_attn
+    inference = Inference(model_name, model_type, proxy, config, args.dataset)
+  File "/home/ubuntu/safetynet/utils/attn_store.py", line 16, in __init__
+    self.manager.load_all()
+  File "/home/ubuntu/safetynet/src/models/model_factory.py", line 124, in load_all
+    self.tokenizer = self.factory.create_tokenizer(self.model_name, self.dataset)
+  File "/home/ubuntu/safetynet/src/models/model_factory.py", line 16, in create_tokenizer
+    tokenizer = AutoTokenizer.from_pretrained(
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/tokenization_auto.py", line 897, in from_pretrained
+    return tokenizer_class.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/tokenization_utils_base.py", line 2271, in from_pretrained
+    return cls._from_pretrained(
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/tokenization_utils_base.py", line 2505, in _from_pretrained
+    tokenizer = cls(*init_inputs, **init_kwargs)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/llama/tokenization_llama_fast.py", line 157, in __init__
+    super().__init__(
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/tokenization_utils_fast.py", line 106, in __init__
+    raise ValueError(
+ValueError: Cannot instantiate this tokenizer from a slow version. If it's based on sentencepiece, make sure you have sentencepiece installed.

logs/anthropic/mistral_attn_normal.log

@@ -0,0 +1,35 @@
+/usr/lib/python3/dist-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.4
+  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
+✅ SafetyNetConfig initialized for mistral
+✅ data_path: safetynet/utils/data/mistral
+Setting up mistral model...
+Completed Anthropic model config for mistral
+Loading model...🦾🔥
+Completed Anthropic model config for mistral
+mistralai/Mistral-7B-Instruct-v0.3
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 196, in _run_module_as_main
+    return _run_code(code, main_globals, None,
+  File "/usr/lib/python3.10/runpy.py", line 86, in _run_code
+    exec(code, run_globals)
+  File "/home/ubuntu/safetynet/utils/attn_store.py", line 171, in <module>
+    results = saving_attn(model_name=args.model,  
+  File "/home/ubuntu/safetynet/utils/attn_store.py", line 90, in saving_attn
+    inference = Inference(model_name, model_type, proxy, config, args.dataset)
+  File "/home/ubuntu/safetynet/utils/attn_store.py", line 16, in __init__
+    self.manager.load_all()
+  File "/home/ubuntu/safetynet/src/models/model_factory.py", line 124, in load_all
+    self.tokenizer = self.factory.create_tokenizer(self.model_name, self.dataset)
+  File "/home/ubuntu/safetynet/src/models/model_factory.py", line 16, in create_tokenizer
+    tokenizer = AutoTokenizer.from_pretrained(
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/auto/tokenization_auto.py", line 897, in from_pretrained
+    return tokenizer_class.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/tokenization_utils_base.py", line 2271, in from_pretrained
+    return cls._from_pretrained(
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/tokenization_utils_base.py", line 2505, in _from_pretrained
+    tokenizer = cls(*init_inputs, **init_kwargs)
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/models/llama/tokenization_llama_fast.py", line 157, in __init__
+    super().__init__(
+  File "/home/ubuntu/.local/lib/python3.10/site-packages/transformers/tokenization_utils_fast.py", line 106, in __init__
+    raise ValueError(
+ValueError: Cannot instantiate this tokenizer from a slow version. If it's based on sentencepiece, make sure you have sentencepiece installed.

logs/anthropic/mistral_beatrix_20260127_181330.log

@@ -0,0 +1,20 @@
+/usr/lib/python3/dist-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.4
+  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
+{'model_name': 'mistral', 'detector': 'beatrix', 'layer_idx': 12, 'model_type': 'backdoor', 'dataset': 'anthropic'}
+Completed Anthropic model config for mistral
+Traceback (most recent call last):
+  File "/usr/lib/python3.10/runpy.py", line 196, in _run_module_as_main
+    return _run_code(code, main_globals, None,
+  File "/usr/lib/python3.10/runpy.py", line 86, in _run_code
+    exec(code, run_globals)
+  File "/home/ubuntu/safetynet/src/analysis/safetynet.py", line 266, in <module>
+    main()
+  File "/home/ubuntu/safetynet/src/analysis/safetynet.py", line 260, in main
+    metrics = monitor.forward()
+  File "/home/ubuntu/safetynet/src/analysis/safetynet.py", line 72, in forward
+    normal_data, harmful_data = self.data_processor.forward()
+  File "/home/ubuntu/safetynet/utils/safetynet/vae_ae_train.py", line 51, in forward
+    normal_attention = self.load_layer_attention("normal")
+  File "/home/ubuntu/safetynet/utils/safetynet/vae_ae_train.py", line 28, in load_layer_attention
+    raise FileNotFoundError(f"Layer directory does not exist: {layer_dir}")
+FileNotFoundError: Layer directory does not exist: safetynet/safetynet/anthropic/mistral/backdoor/normal/layer_12

logs/anthropic/mistral_beatrix_20260127_183051.log

@@ -0,0 +1,29 @@
+/usr/lib/python3/dist-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.4
+  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
+{'model_name': 'mistral', 'detector': 'beatrix', 'layer_idx': 12, 'model_type': 'backdoored', 'dataset': 'anthropic'}
+Completed Anthropic model config for mistral
+Layer Dir: 🦠 safetynet/safetynet/anthropic/mistral/backdoored/normal/layer_12
+Layer Dir: 🦠 safetynet/safetynet/anthropic/mistral/backdoored/harmful/layer_12
+Normal data: torch.Size([3678, 57121]), Harmful data: torch.Size([3335, 57121])
+Training on 2943 normal samples
+Validation: 735 normal, 3335 harmful samples
+🏋🏻‍♀️👟Fitting on training data
+Reducing dims from 57121 to 100
+Feature stats: min=0.000, max=255.875
+/home/ubuntu/safetynet/utils/safetynet/detectors.py:341: RuntimeWarning: overflow encountered in scalar divide
+  delta_j = min((val - max_bounds[j]) / denominator, 10.0)
+✅ Beatrix fitted. Threshold: +0.5033 and -0.2018
+⏭️ Forwarding....
+Reducing dims from 57121 to 100
+/home/ubuntu/safetynet/utils/safetynet/detectors.py:341: RuntimeWarning: overflow encountered in scalar divide
+  delta_j = min((val - max_bounds[j]) / denominator, 10.0)
+⏭️ Forwarding....
+Reducing dims from 57121 to 100
+/home/ubuntu/safetynet/utils/safetynet/detectors.py:341: RuntimeWarning: overflow encountered in scalar divide
+  delta_j = min((val - max_bounds[j]) / denominator, 10.0)
+⏭️ Forwarding....
+Reducing dims from 57121 to 100
+/home/ubuntu/safetynet/utils/safetynet/detectors.py:341: RuntimeWarning: overflow encountered in scalar divide
+  delta_j = min((val - max_bounds[j]) / denominator, 10.0)
+{'auroc': 0.4877111902764215, 'overall_accuracy': 0.4695762981841351, 'overall_precision': 0.5174825174825175, 'overall_recall': 0.022188905547226388, 'overall_f1': 0.0425531914893617, 'train_accuracy': 0.9765545361875637, 'harmful_accuracy': 0.022188905547226388, 'harmful_precision': 1.0, 'harmful_recall': 0.022188905547226388, 'harmful_f1': 0.04341449105309475, 'threshold_lower': 0.18886752426624298, 'threshold_upper': 0.5223360508680344, 'confusion_matrix_overall': array([[5748,  138],
+       [6522,  148]]), 'tp_overall': 148, 'fp_overall': 138, 'tn_overall': 5748, 'fn_overall': 6522}

logs/anthropic/mistral_crow.log

@@ -0,0 +1,71 @@
+/usr/lib/python3/dist-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.4
+  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
+{'model_name': 'mistral', 'model_type': 'backdoored', 'dataset': 'anthropic'}
+Completed Anthropic model config for mistral
+Analyzing layers: [11, 12, 13]
+
+Processing L11 -> L12
+
+/home/ubuntu/.local/lib/python3.10/site-packages/numpy/core/fromnumeric.py:3504: RuntimeWarning: Mean of empty slice.
+  return _methods._mean(a, axis=axis, dtype=dtype,
+/home/ubuntu/.local/lib/python3.10/site-packages/numpy/core/_methods.py:129: RuntimeWarning: invalid value encountered in scalar divide
+  ret = ret.dtype.type(ret / rcount)
+
+/home/ubuntu/safetynet/utils/crow.py:193: RuntimeWarning: Mean of empty slice.
+  self.positive_threshold = train_distances.mean() + self.threshold_scale * train_distances.std()
+/home/ubuntu/.local/lib/python3.10/site-packages/numpy/core/_methods.py:206: RuntimeWarning: Degrees of freedom <= 0 for slice
+  ret = _var(a, axis=axis, dtype=dtype, out=out, ddof=ddof,
+/home/ubuntu/.local/lib/python3.10/site-packages/numpy/core/_methods.py:163: RuntimeWarning: invalid value encountered in divide
+  arrmean = um.true_divide(arrmean, div, out=arrmean,
+/home/ubuntu/.local/lib/python3.10/site-packages/numpy/core/_methods.py:198: RuntimeWarning: invalid value encountered in scalar divide
+  ret = ret.dtype.type(ret / rcount)
+/home/ubuntu/safetynet/utils/crow.py:194: RuntimeWarning: Mean of empty slice.
+  self.negative_threshold = train_distances.mean() - self.threshold_scale * train_distances.std()
+/home/ubuntu/safetynet/utils/crow.py:207: RuntimeWarning: Mean of empty slice.
+  accuracy = (pred_labels == true_labels).mean()
+/home/ubuntu/.local/lib/python3.10/site-packages/numpy/lib/function_base.py:520: RuntimeWarning: Mean of empty slice.
+  avg = a.mean(axis, **keepdims_kw)
+/usr/lib/python3/dist-packages/sklearn/metrics/_classification.py:1221: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.
+  _warn_prf(average, modifier, msg_start, len(result))
+/usr/lib/python3/dist-packages/sklearn/metrics/_classification.py:1221: UndefinedMetricWarning: Recall is ill-defined and being set to 0.0 due to no true samples. Use `zero_division` parameter to control this behavior.
+  _warn_prf(average, modifier, msg_start, len(result))
+/usr/lib/python3/dist-packages/sklearn/metrics/_classification.py:1464: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 due to no true nor predicted samples. Use `zero_division` parameter to control this behavior.
+  _warn_prf(
+==================================================
+🚨 DETECTION RESULTS for HARMFUL DATA of Layer 11 and 12 🚨
+==================================================
+🎯 Accuracy:  nan
+🔍 Precision: 0.0000
+📊 Recall:    0.0000
+⚡ F1-Score:  0.0000
+==================================================
+
+==================================================
+✅ DETECTION RESULTS for NORMAL DATA of Layer 11 and 12 ✅
+==================================================
+🎯 Accuracy:  nan
+🔍 Precision: 0.0000
+📊 Recall:    0.0000
+⚡ F1-Score:  0.0000
+==================================================
+
+Processing L12 -> L13
+
+
+==================================================
+🚨 DETECTION RESULTS for HARMFUL DATA of Layer 12 and 13 🚨
+==================================================
+🎯 Accuracy:  nan
+🔍 Precision: 0.0000
+📊 Recall:    0.0000
+⚡ F1-Score:  0.0000
+==================================================
+
+==================================================
+✅ DETECTION RESULTS for NORMAL DATA of Layer 12 and 13 ✅
+==================================================
+🎯 Accuracy:  nan
+🔍 Precision: 0.0000
+📊 Recall:    0.0000
+⚡ F1-Score:  0.0000
+==================================================

logs/anthropic/mistral_mahalanobis_20260127_182405.log

@@ -0,0 +1,13 @@
+/usr/lib/python3/dist-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.4
+  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
+{'model_name': 'mistral', 'detector': 'mahalanobis', 'layer_idx': 12, 'model_type': 'backdoored', 'dataset': 'anthropic'}
+Completed Anthropic model config for mistral
+Layer Dir: 🦠 safetynet/safetynet/anthropic/mistral/backdoored/normal/layer_12
+Layer Dir: 🦠 safetynet/safetynet/anthropic/mistral/backdoored/harmful/layer_12
+Normal data: torch.Size([3678, 57121]), Harmful data: torch.Size([3335, 57121])
+Training on 2943 normal samples
+Validation: 735 normal, 3335 harmful samples
+
+Fitting Mahalanobis detector...
+{'auroc': 1.0, 'overall_accuracy': 0.14670277158330677, 'overall_precision': 0.23835403726708074, 'overall_recall': 0.27616191904047976, 'overall_f1': 0.2558688706764828, 'train_accuracy': 0.0, 'harmful_accuracy': 0.27616191904047976, 'harmful_precision': 1.0, 'harmful_recall': 0.27616191904047976, 'harmful_f1': 0.43280075187969924, 'threshold_lower': 7117.375, 'threshold_upper': 121680.296875, 'confusion_matrix_overall': array([[   0, 2943],
+       [2414,  921]]), 'tp_overall': 921, 'fp_overall': 2943, 'tn_overall': 0, 'fn_overall': 2414}

logs/anthropic/mistral_pca_20260127_182345.log

@@ -0,0 +1,13 @@
+/usr/lib/python3/dist-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.4
+  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
+{'model_name': 'mistral', 'detector': 'pca', 'layer_idx': 12, 'model_type': 'backdoored', 'dataset': 'anthropic'}
+Completed Anthropic model config for mistral
+Layer Dir: 🦠 safetynet/safetynet/anthropic/mistral/backdoored/normal/layer_12
+Layer Dir: 🦠 safetynet/safetynet/anthropic/mistral/backdoored/harmful/layer_12
+Normal data: torch.Size([3678, 57121]), Harmful data: torch.Size([3335, 57121])
+Training on 2943 normal samples
+Validation: 735 normal, 3335 harmful samples
+
+Fitting PCA detector...
+{'auroc': 0.42101859365933747, 'overall_accuracy': 0.46352341510035044, 'overall_precision': 0.43209876543209874, 'overall_recall': 0.031484257871064465, 'overall_f1': 0.0586920067076579, 'train_accuracy': 0.9531090723751274, 'harmful_accuracy': 0.031484257871064465, 'harmful_precision': 1.0, 'harmful_recall': 0.031484257871064465, 'harmful_f1': 0.06104651162790697, 'threshold_lower': 153.88826370239258, 'threshold_upper': 232.4491081237793, 'confusion_matrix_overall': array([[2805,  138],
+       [3230,  105]]), 'tp_overall': 105, 'fp_overall': 138, 'tn_overall': 2805, 'fn_overall': 3230}

logs/anthropic/mistral_vae_20260127_181336.log

@@ -0,0 +1,2 @@
+/usr/lib/python3/dist-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.4
+  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"

logs/anthropic/mistral_vae_20260127_182312.log

@@ -0,0 +1,28 @@
+/usr/lib/python3/dist-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.4
+  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
+{'model_name': 'mistral', 'detector': 'vae', 'layer_idx': 12, 'model_type': 'backdoored', 'dataset': 'anthropic'}
+Completed Anthropic model config for mistral
+Layer Dir: 🦠 safetynet/safetynet/anthropic/mistral/backdoored/normal/layer_12
+Layer Dir: 🦠 safetynet/safetynet/anthropic/mistral/backdoored/harmful/layer_12
+Normal data: torch.Size([3678, 57121]), Harmful data: torch.Size([3335, 57121])
+Training on 2943 normal samples
+Validation: 735 normal, 3335 harmful samples
+
+Training Mistral-7B-Instruct-v0.3...
+
  0%|                                                                                                       | 0/10 [00:00<?, ?it/s]Epoch 1/10, Loss: 2.4367
+
 10%|█████████▌                                                                                     | 1/10 [00:01<00:11,  1.27s/it]Epoch 2/10, Loss: 1.2650
+
 20%|███████████████████                                                                            | 2/10 [00:02<00:09,  1.18s/it]Epoch 3/10, Loss: 1.2222
+
 30%|████████████████████████████▌                                                                  | 3/10 [00:03<00:08,  1.16s/it]Epoch 4/10, Loss: 1.1849
+
 40%|██████████████████████████████████████                                                         | 4/10 [00:04<00:06,  1.15s/it]Epoch 5/10, Loss: 1.1519
+
 50%|███████████████████████████████████████████████▌                                               | 5/10 [00:05<00:05,  1.14s/it]Epoch 6/10, Loss: 1.1241
+
 60%|█████████████████████████████████████████████████████████                                      | 6/10 [00:06<00:04,  1.13s/it]Epoch 7/10, Loss: 1.1009
+
 70%|██████████████████████████████████████████████████████████████████▌                            | 7/10 [00:08<00:03,  1.13s/it]Epoch 8/10, Loss: 1.0813
+
 80%|████████████████████████████████████████████████████████████████████████████                   | 8/10 [00:09<00:02,  1.13s/it]Epoch 9/10, Loss: 1.0652
+
 90%|█████████████████████████████████████████████████████████████████████████████████████▌         | 9/10 [00:10<00:01,  1.13s/it]Epoch 10/10, Loss: 1.0524
+
+
+Evaluating...
+/home/ubuntu/safetynet/utils/safetynet/vae_ae_train.py:142: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
+  state_dict = torch.load(model_path)
+{'auroc': 0.40981533012901594, 'overall_accuracy': 0.47619047619047616, 'overall_precision': 0.6666666666666666, 'overall_recall': 0.029850746268656716, 'overall_f1': 0.05714285714285715, 'train_accuracy': 0.9830508474576272, 'harmful_accuracy': 0.029850746268656716, 'harmful_precision': 1.0, 'harmful_recall': 0.029850746268656716, 'harmful_f1': 0.05797101449275362, 'threshold_lower': 0.9694035340839391, 'threshold_upper': 1.1496993055449165, 'confusion_matrix_overall': array([[116,   2],
+       [130,   4]]), 'tp_overall': 4, 'fp_overall': 2, 'tn_overall': 116, 'fn_overall': 130}

logs/anthropic/mistral_violin_20260127_184746.log

@@ -0,0 +1,3 @@
+/usr/lib/python3/dist-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.4
+  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
+/usr/bin/python: No module named utils.plot_violin_classification

logs/anthropic/mistral_violin_20260127_184844.log

@@ -0,0 +1,5 @@
+/usr/lib/python3/dist-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.17.3 and <1.25.0 is required for this version of SciPy (detected version 1.26.4
+  warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
+usage: plot_violin_classification.py [-h] --model_name MODEL_NAME --model_type MODEL_TYPE --other_layer_idx OTHER_LAYER_IDX
+                                     --dataset DATASET
+plot_violin_classification.py: error: the following arguments are required: --other_layer_idx