Update dependencies and fix execution count errors

requirements-dev.txt

@@ -4,5 +4,8 @@ datasets
 evaluate
 matplotlib
 numpy
+torchvision
 huggingface-hub
-ipywidgets
+ipywidgets
+scikit-learn
+accelerate

train.ipynb

@@ -182,7 +182,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 9,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -197,7 +197,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 10,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -217,7 +217,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 13,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -238,9 +238,33 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 14,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "You passed `num_labels=10` which is incompatible to the `id2label` map of length `1000`.\n",
+      "Loading weights: 100%|██████████| 200/200 [00:00<00:00, 12072.89it/s]\n",
+      "\u001b[1mViTForImageClassification LOAD REPORT\u001b[0m from: google/vit-base-patch16-224\n",
+      "Key               | Status   |                                                                                           \n",
+      "------------------+----------+-------------------------------------------------------------------------------------------\n",
+      "classifier.weight | MISMATCH | Reinit due to size mismatch - ckpt: torch.Size([1000, 768]) vs model:torch.Size([10, 768])\n",
+      "classifier.bias   | MISMATCH | Reinit due to size mismatch - ckpt: torch.Size([1000]) vs model:torch.Size([10])          \n",
+      "\n",
+      "Notes:\n",
+      "- MISMATCH:\tckpt weights were loaded, but they did not match the original empty weight shapes.\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "num_params = 85,806,346 | trainable_params = 7,690\n"
+     ]
+    }
+   ],
    "source": [
     "model = ViTForImageClassification.from_pretrained(\n",
     "    'google/vit-base-patch16-224',\n",
@@ -269,9 +293,26 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 15,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "ename": "ImportError",
+     "evalue": "Using the `Trainer` with `PyTorch` requires `accelerate>=1.1.0`: Please run `pip install transformers[torch]` or `pip install 'accelerate>=1.1.0'`",
+     "output_type": "error",
+     "traceback": [
+      "\u001b[31m---------------------------------------------------------------------------\u001b[39m",
+      "\u001b[31mImportError\u001b[39m                               Traceback (most recent call last)",
+      "\u001b[36mCell\u001b[39m\u001b[36m \u001b[39m\u001b[32mIn[15]\u001b[39m\u001b[32m, line 1\u001b[39m\n\u001b[32m----> \u001b[39m\u001b[32m1\u001b[39m training_args = TrainingArguments(\n\u001b[32m      2\u001b[39m     output_dir=\u001b[33m\"./cifar10-vit\"\u001b[39m,\n\u001b[32m      3\u001b[39m     per_device_train_batch_size=\u001b[32m16\u001b[39m,\n\u001b[32m      4\u001b[39m     save_strategy=\u001b[33m\"epoch\"\u001b[39m,\n",
+      "\u001b[36mFile \u001b[39m\u001b[32m<string>:112\u001b[39m, in \u001b[36m__create_fn__.<locals>.__init__\u001b[39m\u001b[34m(self, output_dir, per_device_train_batch_size, num_train_epochs, max_steps, learning_rate, lr_scheduler_type, lr_scheduler_kwargs, warmup_steps, optim, optim_args, weight_decay, adam_beta1, adam_beta2, adam_epsilon, optim_target_modules, gradient_accumulation_steps, average_tokens_across_devices, max_grad_norm, label_smoothing_factor, bf16, fp16, bf16_full_eval, fp16_full_eval, tf32, gradient_checkpointing, gradient_checkpointing_kwargs, torch_compile, torch_compile_backend, torch_compile_mode, use_liger_kernel, liger_kernel_config, use_cache, neftune_noise_alpha, torch_empty_cache_steps, auto_find_batch_size, logging_strategy, logging_steps, logging_first_step, log_on_each_node, logging_nan_inf_filter, include_num_input_tokens_seen, log_level, log_level_replica, disable_tqdm, report_to, run_name, project, trackio_space_id, eval_strategy, eval_steps, eval_delay, per_device_eval_batch_size, prediction_loss_only, eval_on_start, eval_do_concat_batches, eval_use_gather_object, eval_accumulation_steps, include_for_metrics, batch_eval_metrics, save_only_model, save_strategy, save_steps, save_on_each_node, save_total_limit, enable_jit_checkpoint, push_to_hub, hub_token, hub_private_repo, hub_model_id, hub_strategy, hub_always_push, hub_revision, load_best_model_at_end, metric_for_best_model, greater_is_better, ignore_data_skip, restore_callback_states_from_checkpoint, full_determinism, seed, data_seed, use_cpu, accelerator_config, parallelism_config, dataloader_drop_last, dataloader_num_workers, dataloader_pin_memory, dataloader_persistent_workers, dataloader_prefetch_factor, remove_unused_columns, label_names, train_sampling_strategy, length_column_name, ddp_find_unused_parameters, ddp_bucket_cap_mb, ddp_broadcast_buffers, ddp_backend, ddp_timeout, fsdp, fsdp_config, deepspeed, debug, skip_memory_metrics, do_train, do_eval, do_predict, resume_from_checkpoint, warmup_ratio, logging_dir, local_rank)\u001b[39m\n",
+      "\u001b[36mFile \u001b[39m\u001b[32m~/repos/computer_vision_classification_model_comparison/.venv/lib/python3.13/site-packages/transformers/training_args.py:1586\u001b[39m, in \u001b[36mTrainingArguments.__post_init__\u001b[39m\u001b[34m(self)\u001b[39m\n\u001b[32m   1584\u001b[39m \u001b[38;5;66;03m# ── 8. Device Init ──\u001b[39;00m\n\u001b[32m   1585\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m is_torch_available():\n\u001b[32m-> \u001b[39m\u001b[32m1586\u001b[39m     \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43mdevice\u001b[49m\n\u001b[32m   1588\u001b[39m \u001b[38;5;66;03m# ── 9. TF32 ──\u001b[39;00m\n\u001b[32m   1589\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m is_torch_available() \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;28mself\u001b[39m.torch_compile:\n",
+      "\u001b[36mFile \u001b[39m\u001b[32m~/repos/computer_vision_classification_model_comparison/.venv/lib/python3.13/site-packages/transformers/training_args.py:1874\u001b[39m, in \u001b[36mTrainingArguments.device\u001b[39m\u001b[34m(self)\u001b[39m\n\u001b[32m   1870\u001b[39m \u001b[38;5;250m\u001b[39m\u001b[33;03m\"\"\"\u001b[39;00m\n\u001b[32m   1871\u001b[39m \u001b[33;03mThe device used by this process.\u001b[39;00m\n\u001b[32m   1872\u001b[39m \u001b[33;03m\"\"\"\u001b[39;00m\n\u001b[32m   1873\u001b[39m requires_backends(\u001b[38;5;28mself\u001b[39m, [\u001b[33m\"\u001b[39m\u001b[33mtorch\u001b[39m\u001b[33m\"\u001b[39m])\n\u001b[32m-> \u001b[39m\u001b[32m1874\u001b[39m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43m_setup_devices\u001b[49m\n",
+      "\u001b[36mFile \u001b[39m\u001b[32m~/.local/share/uv/python/cpython-3.13.2-macos-aarch64-none/lib/python3.13/functools.py:1042\u001b[39m, in \u001b[36mcached_property.__get__\u001b[39m\u001b[34m(self, instance, owner)\u001b[39m\n\u001b[32m   1040\u001b[39m val = cache.get(\u001b[38;5;28mself\u001b[39m.attrname, _NOT_FOUND)\n\u001b[32m   1041\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m val \u001b[38;5;129;01mis\u001b[39;00m _NOT_FOUND:\n\u001b[32m-> \u001b[39m\u001b[32m1042\u001b[39m     val = \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43mfunc\u001b[49m\u001b[43m(\u001b[49m\u001b[43minstance\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m   1043\u001b[39m     \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[32m   1044\u001b[39m         cache[\u001b[38;5;28mself\u001b[39m.attrname] = val\n",
+      "\u001b[36mFile \u001b[39m\u001b[32m~/repos/computer_vision_classification_model_comparison/.venv/lib/python3.13/site-packages/transformers/training_args.py:1765\u001b[39m, in \u001b[36mTrainingArguments._setup_devices\u001b[39m\u001b[34m(self)\u001b[39m\n\u001b[32m   1763\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m is_sagemaker_mp_enabled():\n\u001b[32m   1764\u001b[39m     \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m is_accelerate_available():\n\u001b[32m-> \u001b[39m\u001b[32m1765\u001b[39m         \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mImportError\u001b[39;00m(\n\u001b[32m   1766\u001b[39m             \u001b[33mf\u001b[39m\u001b[33m\"\u001b[39m\u001b[33mUsing the `Trainer` with `PyTorch` requires `accelerate>=\u001b[39m\u001b[38;5;132;01m{\u001b[39;00mACCELERATE_MIN_VERSION\u001b[38;5;132;01m}\u001b[39;00m\u001b[33m`: \u001b[39m\u001b[33m\"\u001b[39m\n\u001b[32m   1767\u001b[39m             \u001b[33mf\u001b[39m\u001b[33m\"\u001b[39m\u001b[33mPlease run `pip install transformers[torch]` or `pip install \u001b[39m\u001b[33m'\u001b[39m\u001b[33maccelerate>=\u001b[39m\u001b[38;5;132;01m{\u001b[39;00mACCELERATE_MIN_VERSION\u001b[38;5;132;01m}\u001b[39;00m\u001b[33m'\u001b[39m\u001b[33m`\u001b[39m\u001b[33m\"\u001b[39m\n\u001b[32m   1768\u001b[39m         )\n\u001b[32m   1769\u001b[39m \u001b[38;5;66;03m# Build kwargs for PartialState; actual init happens below\u001b[39;00m\n\u001b[32m   1770\u001b[39m accelerator_state_kwargs: \u001b[38;5;28mdict\u001b[39m[\u001b[38;5;28mstr\u001b[39m, Any] = {\u001b[33m\"\u001b[39m\u001b[33menabled\u001b[39m\u001b[33m\"\u001b[39m: \u001b[38;5;28;01mTrue\u001b[39;00m, \u001b[33m\"\u001b[39m\u001b[33muse_configured_state\u001b[39m\u001b[33m\"\u001b[39m: \u001b[38;5;28;01mFalse\u001b[39;00m}\n",
+      "\u001b[31mImportError\u001b[39m: Using the `Trainer` with `PyTorch` requires `accelerate>=1.1.0`: Please run `pip install transformers[torch]` or `pip install 'accelerate>=1.1.0'`"
+     ]
+    }
+   ],
    "source": [
     "training_args = TrainingArguments(\n",
     "    output_dir=\"./cifar10-vit\",\n",
@@ -357,9 +398,21 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 18,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "ename": "NameError",
+     "evalue": "name 'training_args' is not defined",
+     "output_type": "error",
+     "traceback": [
+      "\u001b[31m---------------------------------------------------------------------------\u001b[39m",
+      "\u001b[31mNameError\u001b[39m                                 Traceback (most recent call last)",
+      "\u001b[36mCell\u001b[39m\u001b[36m \u001b[39m\u001b[32mIn[18]\u001b[39m\u001b[32m, line 1\u001b[39m\n\u001b[32m----> \u001b[39m\u001b[32m1\u001b[39m model_dir = training_args.output_dir\n\u001b[32m      2\u001b[39m trainer.save_model(model_dir)\n\u001b[32m      3\u001b[39m processor.save_pretrained(model_dir)\n",
+      "\u001b[31mNameError\u001b[39m: name 'training_args' is not defined"
+     ]
+    }
+   ],
    "source": [
     "model_dir = training_args.output_dir\n",
     "trainer.save_model(model_dir)\n",