adjustable workers (#2721)

Faster inference for text and tabular

fastai/learner.py

@@ -223,11 +223,11 @@ def validate(self, ds_idx=1, dl=None, cbs=None):
 
     @delegates(GatherPredsCallback.__init__)
     def get_preds(self, ds_idx=1, dl=None, with_input=False, with_decoded=False, with_loss=False, act=None,
-                  inner=False, reorder=True, cbs=None, **kwargs):
-        if dl is None: dl = self.dls[ds_idx].new(shuffled=False, drop_last=False)
+                  inner=False, reorder=True, cbs=None, n_workers=defaults.cpus, **kwargs):
+        if dl is None: dl = self.dls[ds_idx].new(shuffled=False, drop_last=False, num_workers=n_workers)
         if reorder and hasattr(dl, 'get_idxs'):
             idxs = dl.get_idxs()
-            dl = dl.new(get_idxs = _ConstantFunc(idxs))
+            dl = dl.new(get_idxs = _ConstantFunc(idxs), num_workers=n_workers)
         cb = GatherPredsCallback(with_input=with_input, with_loss=with_loss, **kwargs)
         ctx_mgrs = self.validation_context(cbs=L(cbs)+[cb], inner=inner)
         if with_loss: ctx_mgrs.append(self.loss_not_reduced())

fastai/tabular/learner.py

@@ -11,9 +11,9 @@
 @log_args(but_as=Learner.__init__)
 class TabularLearner(Learner):
     "`Learner` for tabular data"
-    def predict(self, row):
+    def predict(self, row, n_workers=defaults.cpus):
         "Predict on a Pandas Series"
-        dl = self.dls.test_dl(row.to_frame().T)
+        dl = self.dls.test_dl(row.to_frame().T, num_workers=0)
         dl.dataset.conts = dl.dataset.conts.astype(np.float32)
         inp,preds,_,dec_preds = self.get_preds(dl=dl, with_input=True, with_decoded=True)
         b = (*tuplify(inp),*tuplify(dec_preds))

fastai/text/learner.py

@@ -157,7 +157,7 @@ def predict(self, text, n_words=1, no_unk=True, temperature=1., min_p=None, no_b
                 decoder=decode_spec_tokens, only_last_word=False):
         "Return `text` and the `n_words` that come after"
         self.model.reset()
-        idxs = idxs_all = self.dls.test_dl([text]).items[0].to(self.dls.device)
+        idxs = idxs_all = self.dls.test_dl([text], num_workers=0).items[0].to(self.dls.device)
         if no_unk: unk_idx = self.dls.vocab.index(UNK)
         for _ in (range(n_words) if no_bar else progress_bar(range(n_words), leave=False)):
             with self.no_bar(): preds,_ = self.get_preds(dl=[(idxs[None],)])

nbs/13a_learner.ipynb

@@ -426,11 +426,11 @@
     "\n",
     "    @delegates(GatherPredsCallback.__init__)\n",
     "    def get_preds(self, ds_idx=1, dl=None, with_input=False, with_decoded=False, with_loss=False, act=None,\n",
-    "                  inner=False, reorder=True, cbs=None, **kwargs):\n",
-    "        if dl is None: dl = self.dls[ds_idx].new(shuffled=False, drop_last=False)\n",
+    "                  inner=False, reorder=True, cbs=None, n_workers=defaults.cpus, **kwargs):\n",
+    "        if dl is None: dl = self.dls[ds_idx].new(shuffled=False, drop_last=False, num_workers=n_workers)\n",
     "        if reorder and hasattr(dl, 'get_idxs'):\n",
     "            idxs = dl.get_idxs()\n",
-    "            dl = dl.new(get_idxs = _ConstantFunc(idxs))\n",
+    "            dl = dl.new(get_idxs = _ConstantFunc(idxs), num_workers=n_workers)\n",
     "        cb = GatherPredsCallback(with_input=with_input, with_loss=with_loss, **kwargs)\n",
     "        ctx_mgrs = self.validation_context(cbs=L(cbs)+[cb], inner=inner)\n",
     "        if with_loss: ctx_mgrs.append(self.loss_not_reduced())\n",

nbs/37_text.learner.ipynb

@@ -443,7 +443,7 @@
     "                decoder=decode_spec_tokens, only_last_word=False):\n",
     "        \"Return `text` and the `n_words` that come after\"\n",
     "        self.model.reset()\n",
-    "        idxs = idxs_all = self.dls.test_dl([text]).items[0].to(self.dls.device)\n",
+    "        idxs = idxs_all = self.dls.test_dl([text], num_workers=0).items[0].to(self.dls.device)\n",
     "        if no_unk: unk_idx = self.dls.vocab.index(UNK)\n",
     "        for _ in (range(n_words) if no_bar else progress_bar(range(n_words), leave=False)):\n",
     "            with self.no_bar(): preds,_ = self.get_preds(dl=[(idxs[None],)])\n",

nbs/43_tabular.learner.ipynb

@@ -73,9 +73,9 @@
     "@log_args(but_as=Learner.__init__)\n",
     "class TabularLearner(Learner):\n",
     "    \"`Learner` for tabular data\"\n",
-    "    def predict(self, row):\n",
+    "    def predict(self, row, n_workers=defaults.cpus):\n",
     "        \"Predict on a Pandas Series\"\n",
-    "        dl = self.dls.test_dl(row.to_frame().T)\n",
+    "        dl = self.dls.test_dl(row.to_frame().T, num_workers=0)\n",
     "        dl.dataset.conts = dl.dataset.conts.astype(np.float32)\n",
     "        inp,preds,_,dec_preds = self.get_preds(dl=dl, with_input=True, with_decoded=True)\n",
     "        b = (*tuplify(inp),*tuplify(dec_preds))\n",