[no ci] docs update

amaiya · Jun 2, 2023 · ff90bb5 · ff90bb5
1 parent e5aadf1
commit ff90bb5
Show file tree

Hide file tree

Showing 3 changed files with 109 additions and 63 deletions.
diff --git a/docs/core.html b/docs/core.html
@@ -691,15 +691,14 @@ <h1 class="title">Module <code>ktrain.core</code></h1>
         &#34;&#34;&#34;
         ```
         Return numerical estimates of lr using two different methods:
-            1. learning rate associated with minimum numerical gradient
-            2. learning rate associated with minimum loss divided by 10
-        Since neither of these methods are fool-proof and can
+          1. lr associated with minum numerical gradient (None if gradient computation fails)
+          2. lr associated with minimum loss divided by 10
+          3. lr associated with longest valley
+        Since none of these methods are fool-proof and can
         potentially return bad estimates, it is recommended that you
         examine the plot generated by lr_plot to estimate the learning rate.
         Returns:
-          tuple: tuple of the form (float, float), where
-            First element is lr associated with minimum numerical gradient (None if gradient computation fails).
-            Second element is lr associated with minimum loss divided by 10.
+          tuple: tuple of the form (float, float)
         ```
         &#34;&#34;&#34;
         if self.lr_finder is None or not self.lr_finder.find_called():
@@ -4074,15 +4073,14 @@ <h3>Inherited members</h3>
         &#34;&#34;&#34;
         ```
         Return numerical estimates of lr using two different methods:
-            1. learning rate associated with minimum numerical gradient
-            2. learning rate associated with minimum loss divided by 10
-        Since neither of these methods are fool-proof and can
+          1. lr associated with minum numerical gradient (None if gradient computation fails)
+          2. lr associated with minimum loss divided by 10
+          3. lr associated with longest valley
+        Since none of these methods are fool-proof and can
         potentially return bad estimates, it is recommended that you
         examine the plot generated by lr_plot to estimate the learning rate.
         Returns:
-          tuple: tuple of the form (float, float), where
-            First element is lr associated with minimum numerical gradient (None if gradient computation fails).
-            Second element is lr associated with minimum loss divided by 10.
+          tuple: tuple of the form (float, float)
         ```
         &#34;&#34;&#34;
         if self.lr_finder is None or not self.lr_finder.find_called():
@@ -5318,15 +5316,14 @@ <h3>Methods</h3>
 </code></dt>
 <dd>
 <div class="desc"><pre><code>Return numerical estimates of lr using two different methods:
-    1. learning rate associated with minimum numerical gradient
-    2. learning rate associated with minimum loss divided by 10
-Since neither of these methods are fool-proof and can
+  1. lr associated with minum numerical gradient (None if gradient computation fails)
+  2. lr associated with minimum loss divided by 10
+  3. lr associated with longest valley
+Since none of these methods are fool-proof and can
 potentially return bad estimates, it is recommended that you
 examine the plot generated by lr_plot to estimate the learning rate.
 Returns:
-  tuple: tuple of the form (float, float), where
-    First element is lr associated with minimum numerical gradient (None if gradient computation fails).
-    Second element is lr associated with minimum loss divided by 10.
+  tuple: tuple of the form (float, float)
 </code></pre></div>
 <details class="source">
 <summary>
@@ -5336,15 +5333,14 @@ <h3>Methods</h3>
     &#34;&#34;&#34;
     ```
     Return numerical estimates of lr using two different methods:
-        1. learning rate associated with minimum numerical gradient
-        2. learning rate associated with minimum loss divided by 10
-    Since neither of these methods are fool-proof and can
+      1. lr associated with minum numerical gradient (None if gradient computation fails)
+      2. lr associated with minimum loss divided by 10
+      3. lr associated with longest valley
+    Since none of these methods are fool-proof and can
     potentially return bad estimates, it is recommended that you
     examine the plot generated by lr_plot to estimate the learning rate.
     Returns:
-      tuple: tuple of the form (float, float), where
-        First element is lr associated with minimum numerical gradient (None if gradient computation fails).
-        Second element is lr associated with minimum loss divided by 10.
+      tuple: tuple of the form (float, float)
     ```
     &#34;&#34;&#34;
     if self.lr_finder is None or not self.lr_finder.find_called():

diff --git a/docs/text/index.html b/docs/text/index.html
@@ -5977,10 +5977,19 @@ <h3>Methods</h3>
             self.torch_device
         )
 
-    def summarize(self, doc, **kwargs):
+    def summarize(
+        self,
+        doc,
+        max_length=150,
+        min_length=56,
+        no_repeat_ngram_size=3,
+        length_penalty=2.0,
+        num_beams=4,
+        **kwargs,
+    ):
         &#34;&#34;&#34;
         ```
-        summarize document text
+        Summarize document text.  Extra arguments are fed to generate method
         Args:
           doc(str): text of document
         Returns:
@@ -5995,11 +6004,12 @@ <h3>Methods</h3>
             )[&#34;input_ids&#34;].to(self.torch_device)
             summary_ids = self.model.generate(
                 answers_input_ids,
-                num_beams=4,
-                length_penalty=2.0,
-                max_length=142,
-                min_length=56,
-                no_repeat_ngram_size=3,
+                num_beams=num_beams,
+                length_penalty=length_penalty,
+                max_length=max_length,
+                min_length=min_length,
+                no_repeat_ngram_size=no_repeat_ngram_size,
+                **kwargs,
             )
 
             exec_sum = self.tokenizer.decode(
@@ -6014,10 +6024,10 @@ <h3>Ancestors</h3>
 <h3>Methods</h3>
 <dl>
 <dt id="ktrain.text.TransformerSummarizer.summarize"><code class="name flex">
-<span>def <span class="ident">summarize</span></span>(<span>self, doc, **kwargs)</span>
+<span>def <span class="ident">summarize</span></span>(<span>self, doc, max_length=150, min_length=56, no_repeat_ngram_size=3, length_penalty=2.0, num_beams=4, **kwargs)</span>
 </code></dt>
 <dd>
-<div class="desc"><pre><code>summarize document text
+<div class="desc"><pre><code>Summarize document text.  Extra arguments are fed to generate method
 Args:
   doc(str): text of document
 Returns:
@@ -6027,10 +6037,19 @@ <h3>Methods</h3>
 <summary>
 <span>Expand source code</span>
 </summary>
-<pre><code class="python">def summarize(self, doc, **kwargs):
+<pre><code class="python">def summarize(
+    self,
+    doc,
+    max_length=150,
+    min_length=56,
+    no_repeat_ngram_size=3,
+    length_penalty=2.0,
+    num_beams=4,
+    **kwargs,
+):
     &#34;&#34;&#34;
     ```
-    summarize document text
+    Summarize document text.  Extra arguments are fed to generate method
     Args:
       doc(str): text of document
     Returns:
@@ -6045,11 +6064,12 @@ <h3>Methods</h3>
         )[&#34;input_ids&#34;].to(self.torch_device)
         summary_ids = self.model.generate(
             answers_input_ids,
-            num_beams=4,
-            length_penalty=2.0,
-            max_length=142,
-            min_length=56,
-            no_repeat_ngram_size=3,
+            num_beams=num_beams,
+            length_penalty=length_penalty,
+            max_length=max_length,
+            min_length=min_length,
+            no_repeat_ngram_size=no_repeat_ngram_size,
+            **kwargs,
         )
 
         exec_sum = self.tokenizer.decode(

diff --git a/docs/text/summarization/core.html b/docs/text/summarization/core.html
@@ -54,10 +54,19 @@ <h1 class="title">Module <code>ktrain.text.summarization.core</code></h1>
             self.torch_device
         )
 
-    def summarize(self, doc, **kwargs):
+    def summarize(
+        self,
+        doc,
+        max_length=150,
+        min_length=56,
+        no_repeat_ngram_size=3,
+        length_penalty=2.0,
+        num_beams=4,
+        **kwargs,
+    ):
         &#34;&#34;&#34;
         ```
-        summarize document text
+        Summarize document text.  Extra arguments are fed to generate method
         Args:
           doc(str): text of document
         Returns:
@@ -72,11 +81,12 @@ <h1 class="title">Module <code>ktrain.text.summarization.core</code></h1>
             )[&#34;input_ids&#34;].to(self.torch_device)
             summary_ids = self.model.generate(
                 answers_input_ids,
-                num_beams=4,
-                length_penalty=2.0,
-                max_length=142,
-                min_length=56,
-                no_repeat_ngram_size=3,
+                num_beams=num_beams,
+                length_penalty=length_penalty,
+                max_length=max_length,
+                min_length=min_length,
+                no_repeat_ngram_size=no_repeat_ngram_size,
+                **kwargs,
             )
 
             exec_sum = self.tokenizer.decode(
@@ -373,10 +383,19 @@ <h3>Methods</h3>
             self.torch_device
         )
 
-    def summarize(self, doc, **kwargs):
+    def summarize(
+        self,
+        doc,
+        max_length=150,
+        min_length=56,
+        no_repeat_ngram_size=3,
+        length_penalty=2.0,
+        num_beams=4,
+        **kwargs,
+    ):
         &#34;&#34;&#34;
         ```
-        summarize document text
+        Summarize document text.  Extra arguments are fed to generate method
         Args:
           doc(str): text of document
         Returns:
@@ -391,11 +410,12 @@ <h3>Methods</h3>
             )[&#34;input_ids&#34;].to(self.torch_device)
             summary_ids = self.model.generate(
                 answers_input_ids,
-                num_beams=4,
-                length_penalty=2.0,
-                max_length=142,
-                min_length=56,
-                no_repeat_ngram_size=3,
+                num_beams=num_beams,
+                length_penalty=length_penalty,
+                max_length=max_length,
+                min_length=min_length,
+                no_repeat_ngram_size=no_repeat_ngram_size,
+                **kwargs,
             )
 
             exec_sum = self.tokenizer.decode(
@@ -410,10 +430,10 @@ <h3>Ancestors</h3>
 <h3>Methods</h3>
 <dl>
 <dt id="ktrain.text.summarization.core.TransformerSummarizer.summarize"><code class="name flex">
-<span>def <span class="ident">summarize</span></span>(<span>self, doc, **kwargs)</span>
+<span>def <span class="ident">summarize</span></span>(<span>self, doc, max_length=150, min_length=56, no_repeat_ngram_size=3, length_penalty=2.0, num_beams=4, **kwargs)</span>
 </code></dt>
 <dd>
-<div class="desc"><pre><code>summarize document text
+<div class="desc"><pre><code>Summarize document text.  Extra arguments are fed to generate method
 Args:
   doc(str): text of document
 Returns:
@@ -423,10 +443,19 @@ <h3>Methods</h3>
 <summary>
 <span>Expand source code</span>
 </summary>
-<pre><code class="python">def summarize(self, doc, **kwargs):
+<pre><code class="python">def summarize(
+    self,
+    doc,
+    max_length=150,
+    min_length=56,
+    no_repeat_ngram_size=3,
+    length_penalty=2.0,
+    num_beams=4,
+    **kwargs,
+):
     &#34;&#34;&#34;
     ```
-    summarize document text
+    Summarize document text.  Extra arguments are fed to generate method
     Args:
       doc(str): text of document
     Returns:
@@ -441,11 +470,12 @@ <h3>Methods</h3>
         )[&#34;input_ids&#34;].to(self.torch_device)
         summary_ids = self.model.generate(
             answers_input_ids,
-            num_beams=4,
-            length_penalty=2.0,
-            max_length=142,
-            min_length=56,
-            no_repeat_ngram_size=3,
+            num_beams=num_beams,
+            length_penalty=length_penalty,
+            max_length=max_length,
+            min_length=min_length,
+            no_repeat_ngram_size=no_repeat_ngram_size,
+            **kwargs,
         )
 
         exec_sum = self.tokenizer.decode(