Buckets:

hf-doc-build/doc / transformers /v4.30.2 /pt /converting_tensorflow_models.html
rtrm's picture
download
raw
29.9 kB
<meta charset="utf-8" /><meta http-equiv="content-security-policy" content=""><meta name="hf:doc:metadata" content="{&quot;local&quot;:&quot;convertendo-checkpoints-do-tensorflow-para-pytorch&quot;,&quot;sections&quot;:[{&quot;local&quot;:&quot;bert&quot;,&quot;title&quot;:&quot;BERT&quot;},{&quot;local&quot;:&quot;albert&quot;,&quot;title&quot;:&quot;ALBERT&quot;},{&quot;local&quot;:&quot;openai-gpt&quot;,&quot;title&quot;:&quot;OpenAI GPT&quot;},{&quot;local&quot;:&quot;openai-gpt2&quot;,&quot;title&quot;:&quot;OpenAI GPT-2&quot;},{&quot;local&quot;:&quot;transformerxl&quot;,&quot;title&quot;:&quot;Transformer-XL&quot;},{&quot;local&quot;:&quot;xlnet&quot;,&quot;title&quot;:&quot;XLNet&quot;},{&quot;local&quot;:&quot;xlm&quot;,&quot;title&quot;:&quot;XLM&quot;},{&quot;local&quot;:&quot;t5&quot;,&quot;title&quot;:&quot;T5&quot;}],&quot;title&quot;:&quot;Convertendo checkpoints do TensorFlow para Pytorch&quot;}" data-svelte="svelte-1phssyn">
<link rel="modulepreload" href="/docs/transformers/v4.30.2/pt/_app/assets/pages/__layout.svelte-hf-doc-builder.css">
<link rel="modulepreload" href="/docs/transformers/v4.30.2/pt/_app/start-hf-doc-builder.js">
<link rel="modulepreload" href="/docs/transformers/v4.30.2/pt/_app/chunks/vendor-hf-doc-builder.js">
<link rel="modulepreload" href="/docs/transformers/v4.30.2/pt/_app/chunks/paths-hf-doc-builder.js">
<link rel="modulepreload" href="/docs/transformers/v4.30.2/pt/_app/pages/__layout.svelte-hf-doc-builder.js">
<link rel="modulepreload" href="/docs/transformers/v4.30.2/pt/_app/pages/converting_tensorflow_models.mdx-hf-doc-builder.js">
<link rel="modulepreload" href="/docs/transformers/v4.30.2/pt/_app/chunks/Tip-hf-doc-builder.js">
<link rel="modulepreload" href="/docs/transformers/v4.30.2/pt/_app/chunks/IconCopyLink-hf-doc-builder.js">
<link rel="modulepreload" href="/docs/transformers/v4.30.2/pt/_app/chunks/CodeBlock-hf-doc-builder.js">
<h1 class="relative group"><a id="convertendo-checkpoints-do-tensorflow-para-pytorch" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#convertendo-checkpoints-do-tensorflow-para-pytorch"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a>
<span>Convertendo checkpoints do TensorFlow para Pytorch
</span></h1>
<p>Uma interface de linha de comando é fornecida para converter os checkpoints originais Bert/GPT/GPT-2/Transformer-XL/XLNet/XLM em modelos
que podem ser carregados usando os métodos <code>from_pretrained</code> da biblioteca.</p>
<div class="course-tip bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p>A partir da versão 2.3.0 o script de conversão agora faz parte do transformers CLI (<strong>transformers-cli</strong>) disponível em qualquer instalação
transformers &gt;= 2.3.0.</p>
<p>A documentação abaixo reflete o formato do comando <strong>transformers-cli convert</strong>.</p></div>
<h2 class="relative group"><a id="bert" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#bert"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a>
<span>BERT
</span></h2>
<p>Você pode converter qualquer checkpoint do BERT em TensorFlow (em particular <a href="https://github.com/google-research/bert#pre-trained-models" rel="nofollow">os modelos pré-treinados lançados pelo Google</a>) em um arquivo PyTorch usando um
<a href="https://github.com/huggingface/transformers/tree/main/src/transformers/models/bert/convert_bert_original_tf_checkpoint_to_pytorch.py" rel="nofollow">convert_bert_original_tf_checkpoint_to_pytorch.py</a> script.</p>
<p>Esta Interface de Linha de Comando (CLI) recebe como entrada um checkpoint do TensorFlow (três arquivos começando com <code>bert_model.ckpt</code>) e o
arquivo de configuração (<code>bert_config.json</code>), e então cria um modelo PyTorch para esta configuração, carrega os pesos
do checkpoint do TensorFlow no modelo PyTorch e salva o modelo resultante em um arquivo PyTorch que pode
ser importado usando <code>from_pretrained()</code> (veja o exemplo em <a href="quicktour">quicktour</a> , <a href="https://github.com/huggingface/transformers/tree/main/examples/pytorch/text-classification/run_glue.py" rel="nofollow">run_glue.py</a> ).</p>
<p>Você só precisa executar este script de conversão <strong>uma vez</strong> para obter um modelo PyTorch. Você pode então desconsiderar o checkpoint em
TensorFlow (os três arquivos começando com <code>bert_model.ckpt</code>), mas certifique-se de manter o arquivo de configuração (\
<code>bert_config.json</code>) e o arquivo de vocabulário (<code>vocab.txt</code>), pois eles também são necessários para o modelo PyTorch.</p>
<p>Para executar este script de conversão específico, você precisará ter o TensorFlow e o PyTorch instalados (<code>pip install tensorflow</code>). O resto do repositório requer apenas o PyTorch.</p>
<p>Aqui está um exemplo do processo de conversão para um modelo <code>BERT-Base Uncased</code> pré-treinado:</p>
<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
Copied</div></button></div>
<pre><!-- HTML_TAG_START --><span class="hljs-built_in">export</span> BERT_BASE_DIR=/path/to/bert/uncased_L-12_H-768_A-12
transformers-cli convert --model_type bert \
--tf_checkpoint <span class="hljs-variable">$BERT_BASE_DIR</span>/bert_model.ckpt \
--config <span class="hljs-variable">$BERT_BASE_DIR</span>/bert_config.json \
--pytorch_dump_output <span class="hljs-variable">$BERT_BASE_DIR</span>/pytorch_model.bin<!-- HTML_TAG_END --></pre></div>
<p>Você pode baixar os modelos pré-treinados do Google para a conversão <a href="https://github.com/google-research/bert#pre-trained-models" rel="nofollow">aqui</a>.</p>
<h2 class="relative group"><a id="albert" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#albert"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a>
<span>ALBERT
</span></h2>
<p>Converta os checkpoints do modelo ALBERT em TensorFlow para PyTorch usando o
<a href="https://github.com/huggingface/transformers/tree/main/src/transformers/models/albert/convert_albert_original_tf_checkpoint_to_pytorch.py" rel="nofollow">convert_albert_original_tf_checkpoint_to_pytorch.py</a> script.</p>
<p>A Interface de Linha de Comando (CLI) recebe como entrada um checkpoint do TensorFlow (três arquivos começando com <code>model.ckpt-best</code>) e o
arquivo de configuração (<code>albert_config.json</code>), então cria e salva um modelo PyTorch. Para executar esta conversão, você
precisa ter o TensorFlow e o PyTorch instalados.</p>
<p>Aqui está um exemplo do processo de conversão para o modelo <code>ALBERT Base</code> pré-treinado:</p>
<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
Copied</div></button></div>
<pre><!-- HTML_TAG_START --><span class="hljs-built_in">export</span> ALBERT_BASE_DIR=/path/to/albert/albert_base
transformers-cli convert --model_type albert \
--tf_checkpoint <span class="hljs-variable">$ALBERT_BASE_DIR</span>/model.ckpt-best \
--config <span class="hljs-variable">$ALBERT_BASE_DIR</span>/albert_config.json \
--pytorch_dump_output <span class="hljs-variable">$ALBERT_BASE_DIR</span>/pytorch_model.bin<!-- HTML_TAG_END --></pre></div>
<p>Você pode baixar os modelos pré-treinados do Google para a conversão <a href="https://github.com/google-research/albert#pre-trained-models" rel="nofollow">aqui</a>.</p>
<h2 class="relative group"><a id="openai-gpt" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#openai-gpt"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a>
<span>OpenAI GPT
</span></h2>
<p>Aqui está um exemplo do processo de conversão para um modelo OpenAI GPT pré-treinado, supondo que seu checkpoint NumPy
foi salvo com o mesmo formato do modelo pré-treinado OpenAI (veja <a href="https://github.com/openai/finetune-transformer-lm" rel="nofollow">aqui</a>\
)</p>
<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
Copied</div></button></div>
<pre><!-- HTML_TAG_START --><span class="hljs-built_in">export</span> OPENAI_GPT_CHECKPOINT_FOLDER_PATH=/path/to/openai/pretrained/numpy/weights
transformers-cli convert --model_type gpt \
--tf_checkpoint <span class="hljs-variable">$OPENAI_GPT_CHECKPOINT_FOLDER_PATH</span> \
--pytorch_dump_output <span class="hljs-variable">$PYTORCH_DUMP_OUTPUT</span> \
[--config OPENAI_GPT_CONFIG] \
[--finetuning_task_name OPENAI_GPT_FINETUNED_TASK] \<!-- HTML_TAG_END --></pre></div>
<h2 class="relative group"><a id="openai-gpt2" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#openai-gpt2"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a>
<span>OpenAI GPT-2
</span></h2>
<p>Aqui está um exemplo do processo de conversão para um modelo OpenAI GPT-2 pré-treinado (consulte <a href="https://github.com/openai/gpt-2" rel="nofollow">aqui</a>)</p>
<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
Copied</div></button></div>
<pre><!-- HTML_TAG_START --><span class="hljs-built_in">export</span> OPENAI_GPT2_CHECKPOINT_PATH=/path/to/gpt2/pretrained/weights
transformers-cli convert --model_type gpt2 \
--tf_checkpoint <span class="hljs-variable">$OPENAI_GPT2_CHECKPOINT_PATH</span> \
--pytorch_dump_output <span class="hljs-variable">$PYTORCH_DUMP_OUTPUT</span> \
[--config OPENAI_GPT2_CONFIG] \
[--finetuning_task_name OPENAI_GPT2_FINETUNED_TASK]<!-- HTML_TAG_END --></pre></div>
<h2 class="relative group"><a id="transformerxl" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#transformerxl"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a>
<span>Transformer-XL
</span></h2>
<p>Aqui está um exemplo do processo de conversão para um modelo Transformer-XL pré-treinado (consulte <a href="https://github.com/kimiyoung/transformer-xl/tree/master/tf#obtain-and-evaluate-pretrained-modelos-sota" rel="nofollow">aqui</a>)</p>
<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
Copied</div></button></div>
<pre><!-- HTML_TAG_START --><span class="hljs-built_in">export</span> TRANSFO_XL_CHECKPOINT_FOLDER_PATH=/path/to/transfo/xl/checkpoint
transformers-cli convert --model_type transfo_xl \
--tf_checkpoint <span class="hljs-variable">$TRANSFO_XL_CHECKPOINT_FOLDER_PATH</span> \
--pytorch_dump_output <span class="hljs-variable">$PYTORCH_DUMP_OUTPUT</span> \
[--config TRANSFO_XL_CONFIG] \
[--finetuning_task_name TRANSFO_XL_FINETUNED_TASK]<!-- HTML_TAG_END --></pre></div>
<h2 class="relative group"><a id="xlnet" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#xlnet"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a>
<span>XLNet
</span></h2>
<p>Aqui está um exemplo do processo de conversão para um modelo XLNet pré-treinado:</p>
<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
Copied</div></button></div>
<pre><!-- HTML_TAG_START --><span class="hljs-built_in">export</span> TRANSFO_XL_CHECKPOINT_PATH=/path/to/xlnet/checkpoint
<span class="hljs-built_in">export</span> TRANSFO_XL_CONFIG_PATH=/path/to/xlnet/config
transformers-cli convert --model_type xlnet \
--tf_checkpoint <span class="hljs-variable">$TRANSFO_XL_CHECKPOINT_PATH</span> \
--config <span class="hljs-variable">$TRANSFO_XL_CONFIG_PATH</span> \
--pytorch_dump_output <span class="hljs-variable">$PYTORCH_DUMP_OUTPUT</span> \
[--finetuning_task_name XLNET_FINETUNED_TASK] \<!-- HTML_TAG_END --></pre></div>
<h2 class="relative group"><a id="xlm" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#xlm"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a>
<span>XLM
</span></h2>
<p>Aqui está um exemplo do processo de conversão para um modelo XLM pré-treinado:</p>
<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
Copied</div></button></div>
<pre><!-- HTML_TAG_START --><span class="hljs-built_in">export</span> XLM_CHECKPOINT_PATH=/path/to/xlm/checkpoint
transformers-cli convert --model_type xlm \
--tf_checkpoint <span class="hljs-variable">$XLM_CHECKPOINT_PATH</span> \
--pytorch_dump_output <span class="hljs-variable">$PYTORCH_DUMP_OUTPUT</span>
[--config XML_CONFIG] \
[--finetuning_task_name XML_FINETUNED_TASK]<!-- HTML_TAG_END --></pre></div>
<h2 class="relative group"><a id="t5" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#t5"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a>
<span>T5
</span></h2>
<p>Aqui está um exemplo do processo de conversão para um modelo T5 pré-treinado:</p>
<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
Copied</div></button></div>
<pre><!-- HTML_TAG_START --><span class="hljs-built_in">export</span> T5=/path/to/t5/uncased_L-12_H-768_A-12
transformers-cli convert --model_type t5 \
--tf_checkpoint <span class="hljs-variable">$T5</span>/t5_model.ckpt \
--config <span class="hljs-variable">$T5</span>/t5_config.json \
--pytorch_dump_output <span class="hljs-variable">$T5</span>/pytorch_model.bin<!-- HTML_TAG_END --></pre></div>
<script type="module" data-hydrate="1bmkmdn">
import { start } from "/docs/transformers/v4.30.2/pt/_app/start-hf-doc-builder.js";
start({
target: document.querySelector('[data-hydrate="1bmkmdn"]').parentNode,
paths: {"base":"/docs/transformers/v4.30.2/pt","assets":"/docs/transformers/v4.30.2/pt"},
session: {},
route: false,
spa: false,
trailing_slash: "never",
hydrate: {
status: 200,
error: null,
nodes: [
import("/docs/transformers/v4.30.2/pt/_app/pages/__layout.svelte-hf-doc-builder.js"),
import("/docs/transformers/v4.30.2/pt/_app/pages/converting_tensorflow_models.mdx-hf-doc-builder.js")
],
params: {}
}
});
</script>

Xet Storage Details

Size:
29.9 kB
·
Xet hash:
2f47f037f03a4162a1535cee20c9073f91e48843d17d37069a486b4d9542335a

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.