Buckets:
| <meta charset="utf-8" /><meta name="hf:doc:metadata" content="{"title":"Personalizarea componentelor modelului","local":"personalizarea-componentelor-modelului","sections":[{"title":"Clasa attention","local":"clasa-attention","sections":[],"depth":2},{"title":"LoRA","local":"lora","sections":[],"depth":2}],"depth":1}"/> | |
| <link href="/docs/transformers/main/ro/_app/immutable/entry/start.Cj0_j3du.js" rel="modulepreload"> | |
| <link href="/docs/transformers/main/ro/_app/immutable/chunks/CUCuKvpH.js" rel="modulepreload"> | |
| <link href="/docs/transformers/main/ro/_app/immutable/chunks/Dm_uEeBC.js" rel="modulepreload"> | |
| <link href="/docs/transformers/main/ro/_app/immutable/entry/app.2x2TdrVN.js" rel="modulepreload"> | |
| <link href="/docs/transformers/main/ro/_app/immutable/chunks/BFzpZI28.js" rel="modulepreload"> | |
| <link href="/docs/transformers/main/ro/_app/immutable/chunks/DM0RASQS.js" rel="modulepreload"> | |
| <link href="/docs/transformers/main/ro/_app/immutable/chunks/DsnmJJEf.js" rel="modulepreload"> | |
| <link href="/docs/transformers/main/ro/_app/immutable/chunks/DJ5hodCp.js" rel="modulepreload"> | |
| <link href="/docs/transformers/main/ro/_app/immutable/chunks/GYau11Sm.js" rel="modulepreload"> | |
| <link href="/docs/transformers/main/ro/_app/immutable/nodes/0.BonRETMf.js" rel="modulepreload"> | |
| <link href="/docs/transformers/main/ro/_app/immutable/chunks/B1k23FAg.js" rel="modulepreload"> | |
| <link href="/docs/transformers/main/ro/_app/immutable/chunks/utTYfj_T.js" rel="modulepreload"> | |
| <link href="/docs/transformers/main/ro/_app/immutable/nodes/14.DGoAKrjB.js" rel="modulepreload"> | |
| <link href="/docs/transformers/main/ro/_app/immutable/chunks/BcKuJYrM.js" rel="modulepreload"> | |
| <link href="/docs/transformers/main/ro/_app/immutable/chunks/BGhUAj7r.js" rel="modulepreload"> | |
| <!--dxgx02--><meta name="hf:doc:metadata" content="{"title":"Personalizarea componentelor modelului","local":"personalizarea-componentelor-modelului","sections":[{"title":"Clasa attention","local":"clasa-attention","sections":[],"depth":2},{"title":"LoRA","local":"lora","sections":[],"depth":2}],"depth":1}"/><!----> | |
| <link href="/docs/transformers/main/ro/_app/immutable/assets/0.tn0RQdqM.css" rel="modulepreload"> <!--[--><!--[0--><!--[--><!--[0--><!--[--><!--[--><p></p> <div class="items-center shrink-0 min-w-[100px] max-sm:min-w-[50px] justify-end ml-auto flex" style="float: right; margin-left: 10px; display: inline-flex; position: relative; z-index: 10;"><div class="inline-flex rounded-md max-sm:rounded-sm"><button class="inline-flex items-center gap-1 h-7 max-sm:h-7 px-2 max-sm:px-1.5 text-sm font-medium text-gray-800 border border-r-0 rounded-l-md max-sm:rounded-l-sm border-gray-200 bg-white hover:shadow-inner dark:border-gray-850 dark:bg-gray-950 dark:text-gray-200 dark:hover:bg-gray-800" aria-live="polite"><span class="inline-flex items-center justify-center rounded-md p-0.5 max-sm:p-0 hover:text-gray-800 dark:hover:text-gray-200"><svg class="sm:size-3.5 size-3" 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><!----></span> <span>Copy page</span></button> <button class="inline-flex items-center justify-center w-6 max-sm:w-5 h-7 max-sm:h-7 disabled:pointer-events-none text-sm text-gray-500 hover:text-gray-700 dark:hover:text-white rounded-r-md max-sm:rounded-r-sm border border-l transition border-gray-200 bg-white hover:shadow-inner dark:border-gray-850 dark:bg-gray-950 dark:text-gray-200 dark:hover:bg-gray-800" aria-haspopup="menu" aria-expanded="false" aria-label="Open copy menu"><svg class="transition-transform text-gray-400 overflow-visible sm:size-3.5 size-3 rotate-0" width="1em" height="1em" viewBox="0 0 12 7" fill="none" xmlns="http://www.w3.org/2000/svg"><path d="M1 1L6 6L11 1" stroke="currentColor"></path></svg><!----></button></div> <!--[-1--><!--]--></div><!----> <!--[0--><h1 class="relative group"><a id="personalizarea-componentelor-modelului" 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="#personalizarea-componentelor-modelului"><span><svg 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>Personalizarea componentelor modelului</span></h1><!--]--><!----> <p>O altă modalitate de a personaliza un model este să modifici componentele acestuia în loc să scrii un model complet nou, permițându-ți să adaptezi un model la cazul tău specific de utilizare. De exemplu, poți adăuga noi layers sau optimiza mecanismul de attention al unei arhitecturi. Personalizările sunt aplicate direct unui model Transformers, așadar poți continua să folosești funcții precum <code>Trainer</code>, <code>PreTrainedModel</code> și biblioteca <a href="https://huggingface.co/docs/peft/en/index" rel="nofollow">PEFT</a>.</p> <p>Acest ghid îți va arăta cum să personalizezi mecanismul de attention al unui model pentru a-i aplica <a href="https://huggingface.co/docs/peft/conceptual_guides/adapter#low-rank-adaptation-lora" rel="nofollow">Low-Rank Adaptation (LoRA)</a>.</p> <blockquote class="tip"><p>Utilitarul <a href="https://github.com/huggingface/transformers/blob/9985d06add07a4cc691dc54a7e34f54205c04d40/src/transformers/utils/import_utils.py#L2286" rel="nofollow">clear_import_cache</a> este foarte util când modifici și dezvolți iterativ codul modelului. Acesta elimină toate modulele Transformers din cache și permite Python-ului să reîncarce codul modificat fără a reporni constant mediul tău.</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 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 class="language-py "><!----><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoModel | |
| <span class="hljs-keyword">from</span> transformers.utils.import_utils <span class="hljs-keyword">import</span> clear_import_cache | |
| model = AutoModel.from_pretrained(<span class="hljs-string">"bert-base-uncased"</span>) | |
| <span class="hljs-comment"># modificări ale codului modelului</span> | |
| <span class="hljs-comment"># șterge cache-ul pentru a reîncărca codul modificat</span> | |
| clear_import_cache() | |
| <span class="hljs-comment"># re-importă pentru a folosi codul actualizat</span> | |
| model = AutoModel.from_pretrained(<span class="hljs-string">"bert-base-uncased"</span>)<!----></pre></div><!----></blockquote> <!--[1--><h2 class="relative group"><a id="clasa-attention" 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="#clasa-attention"><span><svg 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>Clasa attention</span></h2><!--]--><!----> <p>[Segment Anything] este un model de segmentare a imaginilor care combină proiecția query-key-value (<code>qkv</code>) în mecanismele sale de attention. Pentru a reduce numărul de parametri antrenabili și overhead-ul computațional, poți aplica LoRA proiecției <code>qkv</code>. Aceasta necesită împărțirea proiecției <code>qkv</code> astfel că poți targeta separat <code>q</code> și <code>v</code> cu LoRA.</p> <ol><li>Creează o clasă de attention personalizată, <code>SamVisionAttentionSplit</code>, prin subclasarea clasei originale <code>SamVisionAttention</code>. În <code>__init__</code>, șterge proiecția combinată <code>qkv</code> și creează un layer liniar separat pentru <code>q</code>, <code>k</code> și <code>v</code>.</li></ol> <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 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 class="language-py "><!----><span class="hljs-keyword">import</span> torch | |
| <span class="hljs-keyword">import</span> torch.nn <span class="hljs-keyword">as</span> nn | |
| <span class="hljs-keyword">from</span> transformers.models.sam.modeling_sam <span class="hljs-keyword">import</span> SamVisionAttention | |
| <span class="hljs-keyword">class</span> <span class="hljs-title class_">SamVisionAttentionSplit</span>(SamVisionAttention, nn.Module): | |
| <span class="hljs-keyword">def</span> <span class="hljs-title function_">__init__</span>(<span class="hljs-params">self, config, window_size</span>): | |
| <span class="hljs-built_in">super</span>().__init__(config, window_size) | |
| <span class="hljs-comment"># elimină proiecția combinată qkv</span> | |
| <span class="hljs-keyword">del</span> <span class="hljs-variable language_">self</span>.qkv | |
| <span class="hljs-comment"># proiecții separate q, k, v</span> | |
| <span class="hljs-variable language_">self</span>.q = nn.Linear(config.hidden_size, config.hidden_size, bias=config.qkv_bias) | |
| <span class="hljs-variable language_">self</span>.k = nn.Linear(config.hidden_size, config.hidden_size, bias=config.qkv_bias) | |
| <span class="hljs-variable language_">self</span>.v = nn.Linear(config.hidden_size, config.hidden_size, bias=config.qkv_bias) | |
| <span class="hljs-variable language_">self</span>._register_load_state_dict_pre_hook(<span class="hljs-variable language_">self</span>.split_q_k_v_load_hook)<!----></pre></div><!----> <ol start="2"><li>Funcția <code>_split_qkv_load_hook</code> împarte weights pre-antrenate <code>qkv</code> în weights separate <code>q</code>, <code>k</code> și <code>v</code> la încărcarea modelului pentru a asigura compatibilitatea cu orice model pre-antrenat.</li></ol> <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 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 class="language-py "><!----> <span class="hljs-keyword">def</span> <span class="hljs-title function_">split_q_k_v_load_hook</span>(<span class="hljs-params">self, state_dict, prefix, *args</span>): | |
| keys_to_delete = [] | |
| <span class="hljs-keyword">for</span> key <span class="hljs-keyword">in</span> <span class="hljs-built_in">list</span>(state_dict.keys()): | |
| <span class="hljs-keyword">if</span> <span class="hljs-string">"qkv."</span> <span class="hljs-keyword">in</span> key: | |
| <span class="hljs-comment"># împarte q, k, v din proiecția combinată</span> | |
| q, k, v = state_dict[key].chunk(<span class="hljs-number">3</span>, dim=<span class="hljs-number">0</span>) | |
| <span class="hljs-comment"># înlocuiește cu proiecții individuale q, k, v</span> | |
| state_dict[key.replace(<span class="hljs-string">"qkv."</span>, <span class="hljs-string">"q."</span>)] = q | |
| state_dict[key.replace(<span class="hljs-string">"qkv."</span>, <span class="hljs-string">"k."</span>)] = k | |
| state_dict[key.replace(<span class="hljs-string">"qkv."</span>, <span class="hljs-string">"v."</span>)] = v | |
| <span class="hljs-comment"># marchează vechea cheie qkv pentru ștergere</span> | |
| keys_to_delete.append(key) | |
| <span class="hljs-comment"># elimină cheile qkv vechi</span> | |
| <span class="hljs-keyword">for</span> key <span class="hljs-keyword">in</span> keys_to_delete: | |
| <span class="hljs-keyword">del</span> state_dict[key]<!----></pre></div><!----> <ol start="3"><li>În forward pass, <code>q</code>, <code>k</code> și <code>v</code> sunt calculate separat în timp ce restul mecanismului de attention rămâne același.</li></ol> <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 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 class="language-py "><!----> <span class="hljs-keyword">def</span> <span class="hljs-title function_">forward</span>(<span class="hljs-params">self, hidden_states: torch.Tensor, output_attentions=<span class="hljs-literal">False</span></span>) -> torch.Tensor: | |
| batch_size, height, width, _ = hidden_states.shape | |
| qkv_shapes = (batch_size * <span class="hljs-variable language_">self</span>.num_attention_heads, height * width, -<span class="hljs-number">1</span>) | |
| query = <span class="hljs-variable language_">self</span>.q(hidden_states).reshape((batch_size, height * width,<span class="hljs-variable language_">self</span>.num_attention_heads, -<span class="hljs-number">1</span>)).permute(<span class="hljs-number">0</span>,<span class="hljs-number">2</span>,<span class="hljs-number">1</span>,<span class="hljs-number">3</span>).reshape(qkv_shapes) | |
| key = <span class="hljs-variable language_">self</span>.k(hidden_states).reshape((batch_size, height * width,<span class="hljs-variable language_">self</span>.num_attention_heads, -<span class="hljs-number">1</span>)).permute(<span class="hljs-number">0</span>,<span class="hljs-number">2</span>,<span class="hljs-number">1</span>,<span class="hljs-number">3</span>).reshape(qkv_shapes) | |
| value = <span class="hljs-variable language_">self</span>.v(hidden_states).reshape((batch_size, height * width,<span class="hljs-variable language_">self</span>.num_attention_heads, -<span class="hljs-number">1</span>)).permute(<span class="hljs-number">0</span>,<span class="hljs-number">2</span>,<span class="hljs-number">1</span>,<span class="hljs-number">3</span>).reshape(qkv_shapes) | |
| attn_weights = (query * <span class="hljs-variable language_">self</span>.scale) @ key.transpose(-<span class="hljs-number">2</span>, -<span class="hljs-number">1</span>) | |
| attn_weights = torch.nn.functional.softmax(attn_weights, dtype=torch.float32, dim=-<span class="hljs-number">1</span>).to(query.dtype) | |
| attn_probs = nn.functional.dropout(attn_weights, p=<span class="hljs-variable language_">self</span>.dropout, training=<span class="hljs-variable language_">self</span>.training) | |
| attn_output = (attn_probs @ value).reshape(batch_size, <span class="hljs-variable language_">self</span>.num_attention_heads, height, width, -<span class="hljs-number">1</span>) | |
| attn_output = attn_output.permute(<span class="hljs-number">0</span>, <span class="hljs-number">2</span>, <span class="hljs-number">3</span>, <span class="hljs-number">1</span>, <span class="hljs-number">4</span>).reshape(batch_size, height, width, -<span class="hljs-number">1</span>) | |
| attn_output = <span class="hljs-variable language_">self</span>.proj(attn_output) | |
| <span class="hljs-keyword">if</span> output_attentions: | |
| outputs = (attn_output, attn_weights) | |
| <span class="hljs-keyword">else</span>: | |
| outputs = (attn_output, <span class="hljs-literal">None</span>) | |
| <span class="hljs-keyword">return</span> outputs<!----></pre></div><!----> <p>Atribuie clasa personalizată <code>SamVisionAttentionSplit</code> modulului <code>SamVisionAttention</code> al modelului original pentru a-l înlocui. Toate instanțele <code>SamVisionAttention</code> din model sunt înlocuite cu versiunea de attention împărțită.</p> <p>Încarcă modelul cu <code>from_pretrained()</code>.</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 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 class="language-py "><!----><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> SamModel | |
| <span class="hljs-comment"># încarcă modelul SAM pre-antrenat</span> | |
| model = SamModel.from_pretrained(<span class="hljs-string">"facebook/sam-vit-base"</span>) | |
| <span class="hljs-comment"># înlocuiește clasa de attention în modulul vision_encoder</span> | |
| <span class="hljs-keyword">for</span> layer <span class="hljs-keyword">in</span> model.vision_encoder.layers: | |
| <span class="hljs-keyword">if</span> <span class="hljs-built_in">hasattr</span>(layer, <span class="hljs-string">"attn"</span>): | |
| layer.attn = SamVisionAttentionSplit(model.config.vision_config, model.config.vision_config.window_size)<!----></pre></div><!----> <!--[1--><h2 class="relative group"><a id="lora" 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="#lora"><span><svg 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>LoRA</span></h2><!--]--><!----> <p>Cu proiecții separate <code>q</code>, <code>k</code> și <code>v</code>, aplică LoRA la <code>q</code> și <code>v</code>.</p> <p>Creează un <a href="https://huggingface.co/docs/peft/package_reference/config#peft.PeftConfig" rel="nofollow">LoraConfig</a> și specifică rank-ul <code>r</code>, <code>lora_alpha</code>, <code>lora_dropout</code>, <code>task_type</code> și, cel mai important, modulele de targetat.</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 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 class="language-py "><!----><span class="hljs-keyword">from</span> peft <span class="hljs-keyword">import</span> LoraConfig, get_peft_model | |
| config = LoraConfig( | |
| r=<span class="hljs-number">16</span>, | |
| lora_alpha=<span class="hljs-number">32</span>, | |
| <span class="hljs-comment"># aplică LoRA la q și v</span> | |
| target_modules=[<span class="hljs-string">"q"</span>, <span class="hljs-string">"v"</span>], | |
| lora_dropout=<span class="hljs-number">0.1</span>, | |
| task_type=<span class="hljs-string">"FEATURE_EXTRACTION"</span> | |
| )<!----></pre></div><!----> <p>Pasează modelul și <a href="https://huggingface.co/docs/peft/package_reference/config#peft.PeftConfig" rel="nofollow">LoraConfig</a> la <a href="https://huggingface.co/docs/peft/package_reference/peft_model#peft.get_peft_model" rel="nofollow">get_peft_model</a> pentru a aplica LoRA modelului.</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 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 class="language-py "><!---->model = get_peft_model(model, config)<!----></pre></div><!----> <p>Apelează <a href="https://huggingface.co/docs/peft/package_reference/peft_model#peft.PeftMixedModel.print_trainable_parameters" rel="nofollow">print_trainable_parameters</a> pentru a vizualiza numărul de parametri pe care îi antrenezi ca rezultat față de numărul total de parametri.</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 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 class="language-py "><!---->model.print_trainable_parameters() | |
| <span class="hljs-string">"trainable params: 589,824 || all params: 94,274,096 || trainable%: 0.6256"</span><!----></pre></div><!----> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/transformers/blob/main/docs/source/ro/how_to_hack_models.md" target="_blank"><svg class="mr-1" 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="M31,16l-7,7l-1.41-1.41L28.17,16l-5.58-5.59L24,9l7,7z"></path><path d="M1,16l7-7l1.41,1.41L3.83,16l5.58,5.59L8,23l-7-7z"></path><path d="M12.419,25.484L17.639,6.552l1.932,0.518L14.351,26.002z"></path></svg><!----> <span><span class="underline">Update</span> on GitHub</span></a><!----> <p></p><!--]--><!--]--><!--]--><!--]--><!--]--> <!--[-1--><!--]--><!--]--> | |
| <script> | |
| { | |
| __sveltekit_1piag1k = { | |
| base: "/docs/transformers/main/ro", | |
| assets: "/docs/transformers/main/ro" | |
| }; | |
| const element = document.currentScript.parentElement; | |
| Promise.all([ | |
| import("/docs/transformers/main/ro/_app/immutable/entry/start.Cj0_j3du.js"), | |
| import("/docs/transformers/main/ro/_app/immutable/entry/app.2x2TdrVN.js") | |
| ]).then(([kit, app]) => { | |
| kit.start(app, element, { | |
| node_ids: [0, 14], | |
| data: [null,null], | |
| form: null, | |
| error: null | |
| }); | |
| }); | |
| } | |
| </script> | |
Xet Storage Details
- Size:
- 31.3 kB
- Xet hash:
- 36d7888d7d82b4ae91cd52b82df4a06779b647dab479a666afb86e8c663df641
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.