{
"cells": [
{
"cell_type": "code",
"execution_count": 5,
"id": "55adec97",
"metadata": {},
"outputs": [],
"source": [
"# import torch\n",
"# model_dict = torch.load(\"/home/georg/.cache/huggingface/transformers/f3ea20c1be367dfe97f01cf9a67c400789ee358e4b5865438992eb29efc20ab9.a2f429856af34e4e7fdcd9f1ac6c83e5640d7a553d3acff35d3adcab1ea0e070\")\n",
"# model_dict[\"projector.weight\"] = model_dict[\"classifier.dense.weight\"]\n",
"# model_dict[\"projector.bias\"] = model_dict[\"classifier.dense.bias\"]\n",
"# model_dict[\"classifier.weight\"] = model_dict[\"classifier.output.weight\"]\n",
"# model_dict[\"classifier.bias\"] = model_dict[\"classifier.output.bias\"]\n",
"# del model_dict[\"classifier.dense.weight\"]\n",
"# del model_dict[\"classifier.dense.bias\"]\n",
"# del model_dict[\"classifier.output.weight\"]\n",
"# del model_dict[\"classifier.output.bias\"]\n",
"# torch.save(model_dict, \"/home/georg/.cache/huggingface/transformers/f3ea20c1be367dfe97f01cf9a67c400789ee358e4b5865438992eb29efc20ab9.a2f429856af34e4e7fdcd9f1ac6c83e5640d7a553d3acff35d3adcab1ea0e070\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "61837f17",
"metadata": {},
"outputs": [],
"source": [
"from scipy.special import softmax\n",
"from suno_utils.audio import Audio\n",
"from transformers import AutoConfig, Wav2Vec2FeatureExtractor, Wav2Vec2ForSequenceClassification"
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "52adbe01",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/home/georg/venvs/ml/lib/python3.8/site-packages/transformers/configuration_utils.py:340: UserWarning: Passing `gradient_checkpointing` to a config initialization is deprecated and will be removed in v5 Transformers. Using `model.gradient_checkpointing_enable()` instead, or if you are using the `Trainer` API, pass `gradient_checkpointing=True` in your `TrainingArguments`.\n",
" warnings.warn(\n"
]
}
],
"source": [
"model_name = \"ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition\"\n",
"\n",
"processor = Wav2Vec2FeatureExtractor.from_pretrained(model_name)\n",
"\n",
"config = AutoConfig.from_pretrained(model_name)\n",
"config.classifier_proj_size = 1024\n",
"\n",
"class_labels = [\n",
" label for label, _ in sorted([(k, v) for k, v in config.label2id.items()], key=lambda x: x[-1])\n",
"]\n",
"\n",
"model = Wav2Vec2ForSequenceClassification.from_pretrained(model_name, config=config)"
]
},
{
"cell_type": "code",
"execution_count": 45,
"id": "17243988",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"[('angry', 0.0055537275),\n",
" ('calm', 0.0006214727),\n",
" ('disgust', 0.0013819562),\n",
" ('fearful', 0.0024481819),\n",
" ('happy', 0.009534843),\n",
" ('neutral', 0.0015639199),\n",
" ('sad', 0.00023028407),\n",
" ('surprised', 0.9786654)]"
]
},
"execution_count": 45,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"audio = Audio.from_file(\"ravdess_sample.mp3\").convert(16_000, 2, 1).get_slice(38, 42)\n",
"audio_array = audio.array_float\n",
"input_values = processor(audio_array, sampling_rate=16_000, return_tensors=\"pt\").input_values\n",
"logits = model(input_values).logits[0].detach().numpy()\n",
"probs = softmax(logits)\n",
"list(zip(class_labels, probs))"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "647a6c6e",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "0c55de8a",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "c70e10d7",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": 25,
"id": "baa2d8a9",
"metadata": {},
"outputs": [],
"source": [
"from suno_utils.utils.notebook import *"
]
},
{
"cell_type": "code",
"execution_count": 37,
"id": "90ffc8c3",
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"\n",
" \n",
" "
],
"text/plain": [
""
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# https://www.youtube.com/watch?v=Y7OQoNEu3dY\n",
"audio = Audio.from_file(\"ravdess_sample.mp3\")\n",
"audio.play()"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "2ed88017",
"metadata": {},
"outputs": [],
"source": [
"# neutral 2, 6\n",
"# calm 7, 11\n",
"# happy 12, 16\n",
"# sad 17, 21\n",
"# angry 22, 26\n",
"# fearful 27, 31\n",
"# disgust 32, 37\n",
"# surprise 38, 42"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "39b6912e",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "0c3d830a",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "28261028",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "32603df5",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "80513b2f",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "markdown",
"id": "a6bb9061",
"metadata": {},
"source": [
"## Playground"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "d5fb3894",
"metadata": {},
"outputs": [],
"source": [
"from transformers import pipeline\n",
"\n",
"audio_classifier = pipeline(\n",
" task=\"audio-classification\"\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "49a6dfe9",
"metadata": {},
"outputs": [],
"source": [
"# import torch\n",
"# model_dict = torch.load(\"/home/georg/notebooks/hf_models/pytorch_model.bin\")\n",
"# torch.save(model_dict, \"/home/georg/.cache/huggingface/transformers/f3ea20c1be367dfe97f01cf9a67c400789ee358e4b5865438992eb29efc20ab9.a2f429856af34e4e7fdcd9f1ac6c83e5640d7a553d3acff35d3adcab1ea0e070\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "cf676abb",
"metadata": {},
"outputs": [],
"source": [
"# from transformers import Wav2Vec2FeatureExtractor\n",
"# processor = Wav2Vec2FeatureExtractor.from_pretrained(\"ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition\")"
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "90f5726f",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": 15,
"id": "2608f486",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"16000\n"
]
}
],
"source": [
"import torchaudio\n",
"speech_array, sampling_rate = torchaudio.load(\"/home/georg/data/sample_audio/russia.wav\")\n",
"print(sampling_rate)"
]
},
{
"cell_type": "code",
"execution_count": 19,
"id": "bc5ad361",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"torch.Size([1, 257068])"
]
},
"execution_count": 19,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"speech_array.shape"
]
},
{
"cell_type": "code",
"execution_count": 26,
"id": "a2cccbc3",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": 30,
"id": "a32c75a5",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([0.29815513, 0.02026868, 0.13844891, 0.02443818, 0.40067884,\n",
" 0.02386653, 0.04223155, 0.05191214], dtype=float32)"
]
},
"execution_count": 30,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"probs"
]
},
{
"cell_type": "code",
"execution_count": 37,
"id": "daab92c8",
"metadata": {
"scrolled": false
},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": 36,
"id": "ec579cec",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"['angry', 'calm', 'disgust', 'fearful', 'happy', 'neutral', 'sad', 'surprised']"
]
},
"execution_count": 36,
"metadata": {},
"output_type": "execute_result"
}
],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "5c481ad9",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "e565995c",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "5d9251ab",
"metadata": {},
"outputs": [],
"source": [
"# \tspeech_array, sampling_rate = torchaudio.load(batch[\"path\"])\n",
"# \tbatch[\"speech\"] = resampler(speech_array).squeeze().numpy()\n",
"\n",
"# \tinputs = processor(batch[\"speech\"], sampling_rate=16_000, return_tensors=\"pt\", padding=True)\n",
"\n",
"# \twith torch.no_grad():\n",
"# \t\tlogits = model(inputs.input_values.to(\"cuda\"), attention_mask=inputs.attention_mask.to(\"cuda\")).logits\n",
"\n",
"# \tpred_ids = torch.argmax(logits, dim=-1)\n",
"# \tbatch[\"pred_strings\"] = processor.batch_decode(pred_ids)\n",
"# \treturn batch"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "87de7939",
"metadata": {},
"outputs": [],
"source": [
"# config.hidden_size=1024 # instead of 256\n",
"# config.classifier_proj_size=1024 # instead of 256"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "a53c7179",
"metadata": {},
"outputs": [],
"source": [
"# >>> import torch\n",
"# >>> from transformers import Wav2Vec2FeatureExtractor, Wav2Vec2ForSequenceClassification\n",
"# >>> from datasets import load_dataset\n",
"# >>> processor = Wav2Vec2FeatureExtractor.from_pretrained(\"superb/wav2vec2-base-superb-ks\")\n",
"# >>> model = Wav2Vec2ForSequenceClassification.from_pretrained(\"superb/wav2vec2-base-superb-ks\")\n",
"# >>> ds = load_dataset(\"anton-l/superb_dummy\", \"ks\", split=\"test\")\n",
"# >>> input_values = processor(ds[\"speech\"][4], return_tensors=\"pt\").input_values # Batch size 1\n",
"# >>> logits = model(input_values).logits\n",
"# >>> predicted_class_ids = torch.argmax(logits, dim=-1)\n",
"# >>> # compute loss\n",
"# >>> target_label = \"down\"\n",
"# >>> labels = torch.tensor([model.config.label2id[target_label]])\n",
"# >>> loss = model(input_values, labels=labels).loss"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "412e99cf",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "9283cc3d",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "700c3ad2",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "4afb6276",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "7106c281",
"metadata": {},
"outputs": [],
"source": [
"# ['classifier.output.weight', 'classifier.output.bias', 'classifier.dense.bias', 'classifier.dense.weight']"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "c80805ba",
"metadata": {},
"outputs": [],
"source": [
"# \tspeech_array, sampling_rate = torchaudio.load(batch[\"path\"])\n",
"# \tbatch[\"speech\"] = resampler(speech_array).squeeze().numpy()\n",
"\n",
"# \tinputs = processor(batch[\"speech\"], sampling_rate=16_000, return_tensors=\"pt\", padding=True)\n",
"\n",
"# \twith torch.no_grad():\n",
"# \t\tlogits = model(inputs.input_values.to(\"cuda\"), attention_mask=inputs.attention_mask.to(\"cuda\")).logits\n",
"\n",
"# \tpred_ids = torch.argmax(logits, dim=-1)\n",
"# \tbatch[\"pred_strings\"] = processor.batch_decode(pred_ids)\n",
"# \treturn batch"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "eeca05d3",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "6723a770",
"metadata": {},
"outputs": [],
"source": [
"# import torch\n",
"# import torchaudio\n",
"# from datasets import load_dataset, load_metric\n",
"# from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor\n",
"# import re\n",
"\n",
"# test_dataset = load_dataset(\"common_voice\", \"{lang_id}\", split=\"test\")\n",
"# wer = load_metric(\"wer\")\n",
"\n",
"# processor = Wav2Vec2Processor.from_pretrained(\"{model_id}\")\n",
"# model = Wav2Vec2ForCTC.from_pretrained(\"{model_id}\")\n",
"\n",
"# model.to(\"cuda\")\n",
"\n",
"# chars_to_ignore_regex = '[\\,\\?\\.\\!\\-\\;\\:\\\"\\“]'\n",
"\n",
"# resampler = torchaudio.transforms.Resample(48_000, 16_000)\n",
"\n",
"# # Preprocessing the datasets.\n",
"# # We need to read the aduio files as arrays\n",
"# def speech_file_to_array_fn(batch):\n",
"# \tbatch[\"sentence\"] = re.sub(chars_to_ignore_regex, '', batch[\"sentence\"]).lower()\n",
"# \tspeech_array, sampling_rate = torchaudio.load(batch[\"path\"])\n",
"# \tbatch[\"speech\"] = resampler(speech_array).squeeze().numpy()\n",
"# \treturn batch\n",
"\n",
"# test_dataset = test_dataset.map(speech_file_to_array_fn)\n",
"\n",
"# # Preprocessing the datasets.\n",
"# # We need to read the aduio files as arrays\n",
"# def evaluate(batch):\n",
"# \tinputs = processor(batch[\"speech\"], sampling_rate=16_000, return_tensors=\"pt\", padding=True)\n",
"\n",
"# \twith torch.no_grad():\n",
"# \t\tlogits = model(inputs.input_values.to(\"cuda\"), attention_mask=inputs.attention_mask.to(\"cuda\")).logits\n",
"\n",
"# \tpred_ids = torch.argmax(logits, dim=-1)\n",
"# \tbatch[\"pred_strings\"] = processor.batch_decode(pred_ids)\n",
"# \treturn batch\n",
"\n",
"# result = test_dataset.map(evaluate, batched=True, batch_size=8)\n",
"\n",
"# print(\"WER: {:2f}\".format(100 * wer.compute(predictions=result[\"pred_strings\"], references=result[\"sentence\"])))"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "5a81ce66",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "ac2115bc",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "e2941bc5",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "178ea6db",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "e1f1c7e8",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "69fae354",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "8a23da94",
"metadata": {},
"outputs": [],
"source": [
"/home/georg/.cache/huggingface/transformers/f3ea20c1be367dfe97f01cf9a67c400789ee358e4b5865438992eb29efc20ab9.a2f429856af34e4e7fdcd9f1ac6c83e5640d7a553d3acff35d3adcab1ea0e070\n",
"\n",
"/home/georg/notebooks/hf_models/bk_old_model\n",
"\n",
"/home/georg/notebooks/hf_models/model_renamed_layers"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "60b120d1",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": 3,
"id": "be12b7f5",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"bk_old_model pytorch_model.bin\r\n"
]
}
],
"source": [
"!ls /home/georg/notebooks/hf_models/pytorch_model.bin"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "0638d2c4",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "06e4691b",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": 3,
"id": "f002d019",
"metadata": {},
"outputs": [],
"source": [
"audio_classifier.save_pretrained(\"test\")"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "5dfc8f9a",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/home/georg/venvs/ml/lib/python3.8/site-packages/transformers/configuration_utils.py:340: UserWarning: Passing `gradient_checkpointing` to a config initialization is deprecated and will be removed in v5 Transformers. Using `model.gradient_checkpointing_enable()` instead, or if you are using the `Trainer` API, pass `gradient_checkpointing=True` in your `TrainingArguments`.\n",
" warnings.warn(\n"
]
}
],
"source": [
"from transformers import pipeline\n",
"\n",
"audio_classifier = pipeline(\n",
" task=\"audio-classification\", model=\"test\"\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "6d42e50e",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "11e268e6",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "85a417f6",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "markdown",
"id": "91f4f74a",
"metadata": {},
"source": []
},
{
"cell_type": "code",
"execution_count": 13,
"id": "bdb58c92",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"{\"url\": \"https://huggingface.co/ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition/resolve/main/pytorch_model.bin\", \"etag\": \"\\\"51f2f61d0b1e09d7873ad2ef6a287f99b8e11ad1d2ee44d8278940e952270f03\\\"\"}"
]
}
],
"source": [
"!cat /home/georg/.cache/huggingface/transformers/f3ea20c1be367dfe97f01cf9a67c400789ee358e4b5865438992eb29efc20ab9.a2f429856af34e4e7fdcd9f1ac6c83e5640d7a553d3acff35d3adcab1ea0e070.json"
]
},
{
"cell_type": "code",
"execution_count": 14,
"id": "e046669f",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"--2022-06-02 18:33:17-- https://huggingface.co/ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition/resolve/main/pytorch_model.bin\n",
"Resolving huggingface.co (huggingface.co)... 3.210.158.153, 34.197.58.156, 34.232.89.2, ...\n",
"Connecting to huggingface.co (huggingface.co)|3.210.158.153|:443... connected.\n",
"HTTP request sent, awaiting response... 302 Found\n",
"Location: https://cdn-lfs.huggingface.co/ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition/51f2f61d0b1e09d7873ad2ef6a287f99b8e11ad1d2ee44d8278940e952270f03?response-content-disposition=attachment%3B%20filename%3D%22pytorch_model.bin%22 [following]\n",
"--2022-06-02 18:33:17-- https://cdn-lfs.huggingface.co/ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition/51f2f61d0b1e09d7873ad2ef6a287f99b8e11ad1d2ee44d8278940e952270f03?response-content-disposition=attachment%3B%20filename%3D%22pytorch_model.bin%22\n",
"Resolving cdn-lfs.huggingface.co (cdn-lfs.huggingface.co)... 13.35.73.57, 13.35.73.18, 13.35.73.36, ...\n",
"Connecting to cdn-lfs.huggingface.co (cdn-lfs.huggingface.co)|13.35.73.57|:443... connected.\n",
"HTTP request sent, awaiting response... 200 OK\n",
"Length: 1266155629 (1.2G) [application/octet-stream]\n",
"Saving to: ‘pytorch_model.bin’\n",
"\n",
"pytorch_model.bin 100%[===================>] 1.18G 18.8MB/s in 66s \n",
"\n",
"2022-06-02 18:34:23 (18.4 MB/s) - ‘pytorch_model.bin’ saved [1266155629/1266155629]\n",
"\n"
]
}
],
"source": [
"!wget https://huggingface.co/ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition/resolve/main/pytorch_model.bin"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "da21c7cc",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "8d9280df",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "7442fe0b",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": 5,
"id": "8f956e59",
"metadata": {},
"outputs": [
{
"ename": "KeyError",
"evalue": "'classifier.dense.weight'",
"output_type": "error",
"traceback": [
"\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
"\u001b[0;31mKeyError\u001b[0m Traceback (most recent call last)",
"\u001b[0;32m/tmp/ipykernel_2216167/3297907557.py\u001b[0m in \u001b[0;36m\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mmodel_dict\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m\"classifier.dense.weight\"\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
"\u001b[0;31mKeyError\u001b[0m: 'classifier.dense.weight'"
]
}
],
"source": [
"model_dict[\"classifier.dense.weight\"]"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "c4cdbd20",
"metadata": {},
"outputs": [],
"source": [
"torch.save(model_dict, \"/home/georg/.cache/huggingface/transformers/f3ea20c1be367dfe97f01cf9a67c400789ee358e4b5865438992eb29efc20ab9.a2f429856af34e4e7fdcd9f1ac6c83e5640d7a553d3acff35d3adcab1ea0e070\")"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "528f037b",
"metadata": {},
"outputs": [],
"source": [
"!cp /home/georg/.cache/huggingface/transformers/f3ea20c1be367dfe97f01cf9a67c400789ee358e4b5865438992eb29efc20ab9.a2f429856af34e4e7fdcd9f1ac6c83e5640d7a553d3acff35d3adcab1ea0e070 bk_old_model"
]
},
{
"cell_type": "code",
"execution_count": 25,
"id": "7ad007b1",
"metadata": {},
"outputs": [
{
"ename": "KeyError",
"evalue": "'classifier.weight'",
"output_type": "error",
"traceback": [
"\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
"\u001b[0;31mKeyError\u001b[0m Traceback (most recent call last)",
"\u001b[0;32m/tmp/ipykernel_2149581/3455425435.py\u001b[0m in \u001b[0;36m\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0ma\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m\"classifier.weight\"\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
"\u001b[0;31mKeyError\u001b[0m: 'classifier.weight'"
]
}
],
"source": []
},
{
"cell_type": "code",
"execution_count": 24,
"id": "a3792fd5",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"tensor([[ 0.0138, -0.0051, 0.0004, ..., 0.0172, 0.0315, -0.0207],\n",
" [ 0.0110, -0.0127, -0.0558, ..., 0.0041, 0.0313, 0.0268],\n",
" [-0.0121, 0.0246, -0.0110, ..., 0.0125, -0.0283, 0.0278],\n",
" ...,\n",
" [-0.0034, -0.0121, -0.0134, ..., -0.0138, 0.0046, -0.0177],\n",
" [-0.0245, -0.0233, -0.0110, ..., -0.0033, 0.0048, 0.0309],\n",
" [ 0.0094, -0.0059, -0.0206, ..., -0.0138, -0.0269, -0.0171]],\n",
" device='cuda:0')"
]
},
"execution_count": 24,
"metadata": {},
"output_type": "execute_result"
}
],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "9916f4b6",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "cbd0b2e5",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "77f25eb8",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "0001b9af",
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"id": "892540d7",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.8.10"
}
},
"nbformat": 4,
"nbformat_minor": 5
}