Skip to content

Commit f0c1bd9

Browse files
jackaldenryanjackaldenryan
committed
cleanup
1 parent 6f70ad6 commit f0c1bd9

File tree

4 files changed

+87
-530
lines changed

4 files changed

+87
-530
lines changed

‎.gitignore

Lines changed: 3 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -165,6 +165,8 @@ cython_debug/
165165
# Cache files
166166
cache.db*
167167

168-
169168
# LongMemEval data
170169
longmemeval_data/
170+
171+
# All DS_Store files
172+
.DS_Store

‎tests/evals/data/LongMemEval_mini_dataset_loading.ipynb

Lines changed: 0 additions & 287 deletions
Original file line numberDiff line numberDiff line change
@@ -2113,293 +2113,6 @@
21132113
"# ######## Save to csv\n",
21142114
"# lme_dataset_df_filtered_human_labelling.to_csv(\"lme_dataset_df_filtered_human_labelling.csv\", index=False)"
21152115
]
2116-
},
2117-
{
2118-
"cell_type": "markdown",
2119-
"metadata": {},
2120-
"source": [
2121-
"# Archive"
2122-
]
2123-
},
2124-
{
2125-
"cell_type": "code",
2126-
"execution_count": null,
2127-
"metadata": {},
2128-
"outputs": [],
2129-
"source": [
2130-
"# Deserialize the JSON strings back into lists of dictionaries\n",
2131-
"list1_dicts = json.loads(eval_minidataset_labelled.iloc[0][output_column_name])\n",
2132-
"list2_dicts = json.loads(eval_minidataset_labelled.iloc[1][output_column_name])\n",
2133-
"\n",
2134-
"list3_dicts = json.loads(eval_minidataset_labelled.iloc[6][output_column_name])\n",
2135-
"list4_dicts = json.loads(eval_minidataset_labelled.iloc[7][output_column_name])\n",
2136-
"list5_dicts = json.loads(eval_minidataset_labelled.iloc[12][output_column_name])\n",
2137-
"list6_dicts = json.loads(eval_minidataset_labelled.iloc[13][output_column_name])\n",
2138-
"\n",
2139-
"# Convert the dictionaries back into EntityNode objects\n",
2140-
"list1_nodes = [dict_to_entity(node_dict, EntityNode) for node_dict in list1_dicts]\n",
2141-
"list2_nodes = [dict_to_entity(node_dict, EntityNode) for node_dict in list2_dicts]\n",
2142-
"list3_nodes = [dict_to_entity(node_dict, EntityNode) for node_dict in list3_dicts]\n",
2143-
"list4_nodes = [dict_to_entity(node_dict, EntityNode) for node_dict in list4_dicts]\n",
2144-
"list5_nodes = [dict_to_entity(node_dict, EntityNode) for node_dict in list5_dicts]\n",
2145-
"list6_nodes = [dict_to_entity(node_dict, EntityNode) for node_dict in list6_dicts]\n",
2146-
"\n",
2147-
"# Now list1_nodes, list2_nodes, list3_nodes, and list4_nodes contain the deserialized EntityNode objects\n",
2148-
"\n",
2149-
"#Convert Each List Into A List Of Only The Node Names\n",
2150-
"list1_node_names = [node.name for node in list1_nodes]\n",
2151-
"list2_node_names = [node.name for node in list2_nodes]\n",
2152-
"list3_node_names = [node.name for node in list3_nodes]\n",
2153-
"list4_node_names = [node.name for node in list4_nodes]\n",
2154-
"list5_node_names = [node.name for node in list5_nodes]\n",
2155-
"list6_node_names = [node.name for node in list6_nodes]\n",
2156-
"\n",
2157-
"\n",
2158-
"#Print Out The Lists\n",
2159-
"print(list1_node_names)\n",
2160-
"print(list2_node_names)\n",
2161-
"print(list3_node_names)\n",
2162-
"print(list4_node_names)\n",
2163-
"print(list5_node_names)\n",
2164-
"print(list6_node_names)\n",
2165-
"\n",
2166-
"#NowCollectAndPrintTheNoteSummaries For Each List\n",
2167-
"\n",
2168-
"# Function to print node summaries\n",
2169-
"def print_node_summaries(node_list, list_name):\n",
2170-
" print(f\"Node summaries for {list_name}:\")\n",
2171-
" for node in node_list:\n",
2172-
" # Assuming each node has a 'name' and 'attributes' attribute\n",
2173-
" print(f\"Name: {node.name}, Summary: {node.summary}\")\n",
2174-
" print(\"\\n\")\n",
2175-
"\n",
2176-
"# Print node summaries for each list\n",
2177-
"print_node_summaries(list1_nodes, \"list1\")\n",
2178-
"print_node_summaries(list2_nodes, \"list2\")\n",
2179-
"print_node_summaries(list3_nodes, \"list3\")\n",
2180-
"print_node_summaries(list4_nodes, \"list4\")\n",
2181-
"print_node_summaries(list5_nodes, \"list5\")\n",
2182-
"print_node_summaries(list6_nodes, \"list6\")\n",
2183-
"\n"
2184-
]
2185-
},
2186-
{
2187-
"cell_type": "code",
2188-
"execution_count": 11,
2189-
"metadata": {},
2190-
"outputs": [],
2191-
"source": [
2192-
"# def get_output(row, model):\n",
2193-
"# \"\"\"\n",
2194-
"# Gets the output for a given model\n",
2195-
"# \"\"\"\n",
2196-
"# if row['task_name'] == \"extract_nodes\":\n",
2197-
"# return row['input_extracted_nodes']\n",
2198-
"# elif row['task_name'] == \"dedupe_nodes\":\n",
2199-
"# return row['input_existing_relevant_nodes']\n",
2200-
"# elif row['task_name'] == \"extract_edges\":\n",
2201-
"# return row['input_extracted_edges']\n",
2202-
"# elif row['task_name'] == \"dedupe_edges\":\n",
2203-
"# return row['input_existing_relevant_edges']\n",
2204-
"# elif row['task_name'] == \"extract_edge_dates\":\n",
2205-
"# return row['input_extracted_edge_dates']\n",
2206-
"# elif row['task_name'] == \"edge_invalidation\":\n",
2207-
"# return row['input_edge_invalidation']\n",
2208-
"\n",
2209-
"# def insert_gpt4o_answers(df):\n",
2210-
"# \"\"\"\n",
2211-
"# Inserts gpt4o answers by doing ingestion in the right order and filling extra input columns as needed\n",
2212-
"# \"\"\"\n",
2213-
"# for _, row in df.iterrows():\n",
2214-
"# # Get the output\n",
2215-
"# output_gpt4o = get_output(row, \"gpt4o\")"
2216-
]
2217-
},
2218-
{
2219-
"cell_type": "code",
2220-
"execution_count": 12,
2221-
"metadata": {},
2222-
"outputs": [],
2223-
"source": [
2224-
"# ######## Write a multisession to a txt file\n",
2225-
"# question_id = \"gpt4_93159ced_abs\"\n",
2226-
"# multi_session = lme_dataset_df[lme_dataset_df['question_id'] == question_id][\"haystack_sessions\"].iloc[0]\n",
2227-
"\n",
2228-
"\n",
2229-
"# with open(f'{question_id}.txt', 'w') as f:\n",
2230-
"\n",
2231-
"\n",
2232-
"# for session in multi_session:\n",
2233-
"# f.write(\"New session \" + \"*\"*200 + \"\\n\")\n",
2234-
"# for message in session:\n",
2235-
" \n",
2236-
"# f.write(f\"{message}\\n\")\n",
2237-
"# f.write(\"\\n\")"
2238-
]
2239-
},
2240-
{
2241-
"cell_type": "code",
2242-
"execution_count": 13,
2243-
"metadata": {},
2244-
"outputs": [],
2245-
"source": [
2246-
"# ######## Method to save all of the snippets (or only firsts/lasts) of the specified multi-sessions to a CSV file\n",
2247-
"\n",
2248-
"\n",
2249-
"# def make_messages_readable(messages):\n",
2250-
"# if len(messages) == 0:\n",
2251-
"# return []\n",
2252-
"# if messages == [None]:\n",
2253-
"# return None\n",
2254-
"# result_string = \"\"\n",
2255-
"# for message in messages:\n",
2256-
"# if message is None:\n",
2257-
"# continue\n",
2258-
"# result_string += \"|\"*80 + f\" {message['role']} \" + \"|\"*80 + \"\\n\\n\" + f\"{message['content']}\\n\\n\"\n",
2259-
"# return result_string\n",
2260-
"\n",
2261-
"\n",
2262-
"\n",
2263-
"# def handle_message_readability(snippet, readableMessages, spreadOutPreviousMessages, max_num_previous_messages):\n",
2264-
"# if readableMessages:\n",
2265-
"# snippet['message'] = make_messages_readable([snippet['message']])\n",
2266-
"# if spreadOutPreviousMessages:\n",
2267-
"# for i in range(max_num_previous_messages):\n",
2268-
"# snippet[f\"previous_message_{i+1}\"] = make_messages_readable([snippet[f\"previous_message_{i+1}\"]])\n",
2269-
"# return snippet\n",
2270-
"# snippet['previous_messages'] = make_messages_readable(snippet['previous_messages'])\n",
2271-
"# return snippet\n",
2272-
" \n",
2273-
"# snippet['message'] = json.dumps(snippet['message'])\n",
2274-
"# if spreadOutPreviousMessages:\n",
2275-
"# for i in range(max_num_previous_messages):\n",
2276-
"# snippet[f\"previous_message_{i+1}\"] = json.dumps(snippet[f\"previous_message_{i+1}\"])\n",
2277-
"# return snippet\n",
2278-
"# snippet['previous_messages'] = json.dumps(snippet['previous_messages'])\n",
2279-
"# return snippet\n",
2280-
"\n",
2281-
"# def save_multi_session_snippets_to_csv(question_ids, max_num_previous_messages=5, mode=\"all\", readableMessages=False, spreadOutPreviousMessages=False, snippet_index=None):\n",
2282-
"# \"\"\"\n",
2283-
"# Creates a csv where each row is a \"snippet\" from longmemeval. A snippet is a message and set of previous messages.\n",
2284-
"\n",
2285-
"# mode:\n",
2286-
"# mode=\"all\" --> saves all possible snippets for the specified question_ids. \n",
2287-
"# mode=\"firsts_only\" --> saves only the first snippet for each question_id.\n",
2288-
"# mode=\"lasts_only\" --> saves only the last snippet for each question_id.\n",
2289-
"# mode=\"index_only\" --> saves only the snippet at snippet_index for each question_id.\n",
2290-
"\n",
2291-
"# readableMessages:\n",
2292-
"# readableMessages=True --> saves the messages in a readable format, intended for reading in Google sheets.\n",
2293-
"# readableMessages=False --> saves the messages in a json format.\n",
2294-
"\n",
2295-
"# spreadOutPreviousMessages:\n",
2296-
"# spreadOutPreviousMessages=True --> spreads out the previous messages across multiple columns.\n",
2297-
"# spreadOutPreviousMessages=False --> saves all previous messages in a single column.\n",
2298-
" \n",
2299-
"# \"\"\"\n",
2300-
"\n",
2301-
"# all_snippets = []\n",
2302-
"# indices = []\n",
2303-
"# for question_id in question_ids:\n",
2304-
" \n",
2305-
"# ######## First, lets combine the sessions and dates into a list of dicts\n",
2306-
"# row = lme_dataset_df[lme_dataset_df['question_id'] == question_id]\n",
2307-
"\n",
2308-
"# if row.empty:\n",
2309-
"# raise ValueError(f\"No question found with ID: {question_id}\")\n",
2310-
"\n",
2311-
"\n",
2312-
"# # Extract the haystack_sessions column value\n",
2313-
"# sessions = row['haystack_sessions'].iloc[0]\n",
2314-
"\n",
2315-
"# # Get the haystack_dates column value\n",
2316-
"# session_dates = row['haystack_dates'].iloc[0]\n",
2317-
"\n",
2318-
"# # Combine into list of dictionaries\n",
2319-
"# session_and_dates = [\n",
2320-
"# {\n",
2321-
"# \"session\": session,\n",
2322-
"# \"date\": datetime.strptime(date, \"%Y/%m/%d (%a) %H:%M\")\n",
2323-
"# } \n",
2324-
"# for session, date in zip(sessions, session_dates)\n",
2325-
"# ]\n",
2326-
"\n",
2327-
"# # Sort by date from earliest to latest\n",
2328-
"# session_and_dates.sort(key=lambda x: x[\"date\"])\n",
2329-
"\n",
2330-
"\n",
2331-
"# all_snippets_this_session = []\n",
2332-
"\n",
2333-
"# total_num_messages = sum([len(session_and_date[\"session\"]) for session_and_date in session_and_dates])\n",
2334-
"\n",
2335-
"# message_index_across_sessions = 0\n",
2336-
"# for session_index, session_and_date in enumerate(session_and_dates):\n",
2337-
"# for message_index_within_session, message in enumerate(session_and_date[\"session\"]):\n",
2338-
" \n",
2339-
"# num_previous_messages = min(max_num_previous_messages, message_index_across_sessions)\n",
2340-
"# previous_snippets = all_snippets_this_session[message_index_across_sessions-num_previous_messages:]\n",
2341-
"# previous_messages_only = [previous_snippet[\"message\"] for previous_snippet in previous_snippets]\n",
2342-
"\n",
2343-
"# snippet = {\n",
2344-
"# \"question_id\": question_id,\n",
2345-
"# \"multisession_index\": int(row.index[0]),\n",
2346-
"# \"session_index\": session_index,\n",
2347-
"# \"message_index_within_session\": message_index_within_session,\n",
2348-
"# \"session_date\": session_and_date[\"date\"],\n",
2349-
"# }\n",
2350-
"\n",
2351-
"# if spreadOutPreviousMessages:\n",
2352-
"# previous_messages_only_padded = previous_messages_only + [None] * (max_num_previous_messages - len(previous_messages_only))\n",
2353-
"# for i, prev_msg in enumerate(previous_messages_only_padded):\n",
2354-
"# snippet[f\"previous_message_{i+1}\"] = prev_msg\n",
2355-
"# snippet[\"message\"] = message\n",
2356-
"# else:\n",
2357-
"# snippet[\"message\"] = message\n",
2358-
"# snippet[\"previous_messages\"] = previous_messages_only\n",
2359-
"\n",
2360-
"# all_snippets_this_session.append(snippet)\n",
2361-
"# message_index_across_sessions += 1\n",
2362-
"\n",
2363-
"# if mode == \"firsts_only\":\n",
2364-
"# all_snippets_this_session = [all_snippets_this_session[0]]\n",
2365-
"# elif mode == \"lasts_only\":\n",
2366-
"# all_snippets_this_session = [all_snippets_this_session[-1]]\n",
2367-
"# elif mode == \"index_only\":\n",
2368-
"# all_snippets_this_session = [all_snippets_this_session[snippet_index]]\n",
2369-
"\n",
2370-
"# all_snippets.extend(all_snippets_this_session)\n",
2371-
"# indices.append(int(row.index[0]))\n",
2372-
"\n",
2373-
"\n",
2374-
"# filename = \"lme_samples_indices=\"\n",
2375-
"# indices.sort() \n",
2376-
"# num_indices = len(indices)\n",
2377-
"# if num_indices < 4:\n",
2378-
"# for index in indices:\n",
2379-
"# filename += f\"_{index}\"\n",
2380-
"# else:\n",
2381-
"# filename += f\"_{indices[0]}_etc_{indices[-1]}\"\n",
2382-
"\n",
2383-
"# if mode == \"firsts_only\":\n",
2384-
"# filename += \"_firsts_only\"\n",
2385-
"# elif mode == \"lasts_only\":\n",
2386-
"# filename += \"_lasts_only\"\n",
2387-
"# elif mode == \"index_only\":\n",
2388-
"# filename += f\"_index_only={snippet_index}\"\n",
2389-
"# if spreadOutPreviousMessages:\n",
2390-
"# filename += \"_spreadOutPreviousMessages\"\n",
2391-
"# if readableMessages:\n",
2392-
"# filename += \"_readable\"\n",
2393-
"# filename += \".csv\"\n",
2394-
" \n",
2395-
" \n",
2396-
"# with open(filename, \"w\", newline=\"\") as csvfile:\n",
2397-
"# writer = csv.DictWriter(csvfile, fieldnames=all_snippets[0].keys())\n",
2398-
"# writer.writeheader()\n",
2399-
"# for snippet in all_snippets:\n",
2400-
"# processed_snippet = handle_message_readability(snippet, readableMessages, spreadOutPreviousMessages, max_num_previous_messages)\n",
2401-
"# writer.writerow(processed_snippet)"
2402-
]
24032116
}
24042117
],
24052118
"metadata": {

‎tests/evals/data/utils.py

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -97,6 +97,8 @@ async def ingest_and_label_snippet(llm_client, snippet_df, output_column_name):
9797
)
9898
snippet_df.at[extract_edges_row.index[0], output_column_name] = json.dumps([entity_to_dict(edge) for edge in extracted_edges])
9999

100+
####### TODO: finish implementing below
101+
100102
#### Process 'dedupe_edges' task
101103
# dedupe_edges_row = message_df[message_df['task_name'] == 'dedupe_edges']
102104
# assert len(dedupe_edges_row) == 1, "There should be exactly one row for 'dedupe_edges'"

0 commit comments

Comments
 (0)