Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

min_len arg added to make_batch fun #140

Open
wants to merge 1 commit into
base: 2023
Choose a base branch
from
Open

min_len arg added to make_batch fun #140

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
27 changes: 15 additions & 12 deletions week02_classification/seminar.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -301,13 +301,16 @@
"source": [
"UNK_IX, PAD_IX = map(token_to_id.get, [UNK, PAD])\n",
"\n",
"def as_matrix(sequences, max_len=None):\n",
"def as_matrix(sequences, max_len=None, min_len=None):\n",
" \"\"\" Convert a list of tokens into a matrix with padding \"\"\"\n",
" if isinstance(sequences[0], str):\n",
" sequences = list(map(str.split, sequences))\n",
" \n",
" max_len = min(max(map(len, sequences)), max_len or float('inf'))\n",
" \n",
"\n",
" if min_len is not None and min_len > max_len:\n",
" max_len = min_len\n",
" \n",
" matrix = np.full((len(sequences), max_len), np.int32(PAD_IX))\n",
" for i,seq in enumerate(sequences):\n",
" row_ix = [token_to_id.get(word, UNK_IX) for word in seq[:max_len]]\n",
Expand Down Expand Up @@ -407,14 +410,14 @@
" return batch_tensors\n",
"\n",
"\n",
"def make_batch(data, max_len=None, word_dropout=0, device=device):\n",
"def make_batch(data, max_len=None, min_len=None, word_dropout=0, device=device):\n",
" \"\"\"\n",
" Creates a keras-friendly dict from the batch data.\n",
" :param word_dropout: replaces token index with UNK_IX with this probability\n",
" :returns: a dict with {'title' : int64[batch, title_max_len]\n",
" \"\"\"\n",
" batch = {}\n",
" batch[\"Title\"] = as_matrix(data[\"Title\"].values, max_len)\n",
" batch[\"Title\"] = as_matrix(data[\"Title\"].values, max_len, min_len)\n",
" batch[\"FullDescription\"] = as_matrix(data[\"FullDescription\"].values, max_len)\n",
" batch['Categorical'] = categorical_vectorizer.transform(data[categorical_columns].apply(dict, axis=1))\n",
" \n",
Expand Down Expand Up @@ -627,7 +630,7 @@
"metadata": {},
"outputs": [],
"source": [
"def explain(model, sample, col_name='Title'):\n",
"def explain(model, sample, col_name='Title', min_len=None):\n",
" \"\"\" Computes the effect each word had on model predictions \"\"\"\n",
" sample = dict(sample)\n",
" sample_col_tokens = [tokens[token_to_id.get(tok, 0)] for tok in sample[col_name].split()]\n",
Expand All @@ -637,7 +640,7 @@
" data_drop_one_token.loc[drop_i, col_name] = ' '.join(UNK if i == drop_i else tok\n",
" for i, tok in enumerate(sample_col_tokens)) \n",
"\n",
" *predictions_drop_one_token, baseline_pred = model(make_batch(data_drop_one_token, device=device)).detach().cpu()\n",
" *predictions_drop_one_token, baseline_pred = model(make_batch(data_drop_one_token, min_len=min_len, device=device)).detach().cpu()\n",
" diffs = baseline_pred - torch.Tensor(predictions_drop_one_token)\n",
" return list(zip(sample_col_tokens, diffs))"
]
Expand Down Expand Up @@ -681,10 +684,10 @@
"outputs": [],
"source": [
"i = 36605\n",
"tokens_and_weights = explain(model, data.loc[i], \"Title\")\n",
"tokens_and_weights = explain(model, data.loc[i], col_name=\"Title\", min_len=3)\n",
"draw_html([(tok, weight * 5) for tok, weight in tokens_and_weights], font_style='font-size:20px;');\n",
"\n",
"tokens_and_weights = explain(model, data.loc[i], \"FullDescription\")\n",
"tokens_and_weights = explain(model, data.loc[i], col_name=\"FullDescription\", min_len=3)\n",
"draw_html([(tok, weight * 10) for tok, weight in tokens_and_weights]);"
]
},
Expand All @@ -695,10 +698,10 @@
"outputs": [],
"source": [
"i = 12077\n",
"tokens_and_weights = explain(model, data.loc[i], \"Title\")\n",
"tokens_and_weights = explain(model, data.loc[i], col_name=\"Title\", min_len=3)\n",
"draw_html([(tok, weight * 5) for tok, weight in tokens_and_weights], font_style='font-size:20px;');\n",
"\n",
"tokens_and_weights = explain(model, data.loc[i], \"FullDescription\")\n",
"tokens_and_weights = explain(model, data.loc[i], col_name=\"FullDescription\", min_len=3)\n",
"draw_html([(tok, weight * 10) for tok, weight in tokens_and_weights]);"
]
},
Expand All @@ -712,10 +715,10 @@
"print(\"Index:\", i)\n",
"print(\"Salary (gbp):\", np.expm1(model(make_batch(data.iloc[i: i+1], device=device)).detach().cpu()))\n",
"\n",
"tokens_and_weights = explain(model, data.loc[i], \"Title\")\n",
"tokens_and_weights = explain(model, data.loc[i], col_name=\"Title\", min_len=3)\n",
"draw_html([(tok, weight * 5) for tok, weight in tokens_and_weights], font_style='font-size:20px;');\n",
"\n",
"tokens_and_weights = explain(model, data.loc[i], \"FullDescription\")\n",
"tokens_and_weights = explain(model, data.loc[i], col_name=\"FullDescription\", min_len=3)\n",
"draw_html([(tok, weight * 10) for tok, weight in tokens_and_weights]);"
]
},
Expand Down