instructions, more
Browse files- README.md +24 -3
- match.ipynb +40 -18
README.md
CHANGED
@@ -95,7 +95,28 @@ configs:
|
|
95 |
|
96 |
Sourced from https://github.com/mosaicml/llm-foundry/blob/main/scripts/eval/local_data/world_knowledge/jeopardy_all.jsonl
|
97 |
|
98 |
-
Description: Jeopardy consists of 2,117 Jeopardy questions separated into 5 categories:
|
99 |
-
Literature, American History, World History, Word Origins, and Science. The model is expected
|
100 |
-
to give the exact correct response to the question. It was custom curated by MosaicML from a
|
101 |
larger Jeopardy set available on [Huggingface](https://huggingface.co/datasets/jeopardy).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
95 |
|
96 |
Sourced from https://github.com/mosaicml/llm-foundry/blob/main/scripts/eval/local_data/world_knowledge/jeopardy_all.jsonl
|
97 |
|
98 |
+
Description: Jeopardy consists of 2,117 Jeopardy questions separated into 5 categories:
|
99 |
+
Literature, American History, World History, Word Origins, and Science. The model is expected
|
100 |
+
to give the exact correct response to the question. It was custom curated by MosaicML from a
|
101 |
larger Jeopardy set available on [Huggingface](https://huggingface.co/datasets/jeopardy).
|
102 |
+
|
103 |
+
## How to use
|
104 |
+
|
105 |
+
```python
|
106 |
+
from datasets import load_dataset
|
107 |
+
|
108 |
+
dataset = load_dataset("soldni/jeopardy", "mosaicml_gauntlet")
|
109 |
+
model = ...
|
110 |
+
tokenizer = ...
|
111 |
+
|
112 |
+
# Given context, try to predict the continuation
|
113 |
+
for row in dataset:
|
114 |
+
input_ids = tokenizer(row['context'], return_tensors='pt').to(model.device)
|
115 |
+
outputs = model.generate(input_ids, max_new_tokens=100)
|
116 |
+
decoded = tokenizer.decode(outputs[0], skip_special_tokens=True)
|
117 |
+
correct = row['continuation'] in decoded
|
118 |
+
print("Gold:", row['continuation'])
|
119 |
+
print("Pred:", decoded)
|
120 |
+
print("Correct?", correct)
|
121 |
+
print("----")
|
122 |
+
```
|
match.ipynb
CHANGED
@@ -258,14 +258,16 @@
|
|
258 |
},
|
259 |
{
|
260 |
"cell_type": "code",
|
261 |
-
"execution_count":
|
262 |
"metadata": {},
|
263 |
"outputs": [
|
264 |
{
|
265 |
"name": "stdout",
|
266 |
"output_type": "stream",
|
267 |
"text": [
|
268 |
-
"Missing: 0\n"
|
|
|
|
|
269 |
]
|
270 |
}
|
271 |
],
|
@@ -275,6 +277,8 @@
|
|
275 |
"\n",
|
276 |
"missing = [{**v, 'key': k} for k, v in new_questions.items() if k not in old_questions_subset]\n",
|
277 |
"print(f\"Missing: {len(missing)}\")\n",
|
|
|
|
|
278 |
"for m in missing:\n",
|
279 |
" print(m['context'])\n",
|
280 |
" print(m['continuation'])\n",
|
@@ -291,13 +295,21 @@
|
|
291 |
},
|
292 |
{
|
293 |
"cell_type": "code",
|
294 |
-
"execution_count":
|
295 |
"metadata": {},
|
296 |
"outputs": [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
297 |
{
|
298 |
"data": {
|
299 |
"application/vnd.jupyter.widget-view+json": {
|
300 |
-
"model_id": "
|
301 |
"version_major": 2,
|
302 |
"version_minor": 0
|
303 |
},
|
@@ -311,28 +323,21 @@
|
|
311 |
{
|
312 |
"data": {
|
313 |
"application/vnd.jupyter.widget-view+json": {
|
314 |
-
"model_id": "
|
315 |
"version_major": 2,
|
316 |
"version_minor": 0
|
317 |
},
|
318 |
"text/plain": [
|
319 |
-
"Creating parquet from Arrow format:
|
320 |
]
|
321 |
},
|
322 |
"metadata": {},
|
323 |
"output_type": "display_data"
|
324 |
},
|
325 |
-
{
|
326 |
-
"name": "stderr",
|
327 |
-
"output_type": "stream",
|
328 |
-
"text": [
|
329 |
-
"No files have been modified since last commit. Skipping to prevent empty commit.\n"
|
330 |
-
]
|
331 |
-
},
|
332 |
{
|
333 |
"data": {
|
334 |
"application/vnd.jupyter.widget-view+json": {
|
335 |
-
"model_id": "
|
336 |
"version_major": 2,
|
337 |
"version_minor": 0
|
338 |
},
|
@@ -346,7 +351,7 @@
|
|
346 |
{
|
347 |
"data": {
|
348 |
"application/vnd.jupyter.widget-view+json": {
|
349 |
-
"model_id": "
|
350 |
"version_major": 2,
|
351 |
"version_minor": 0
|
352 |
},
|
@@ -357,13 +362,27 @@
|
|
357 |
"metadata": {},
|
358 |
"output_type": "display_data"
|
359 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
360 |
{
|
361 |
"data": {
|
362 |
"text/plain": [
|
363 |
-
"CommitInfo(commit_url='https://huggingface.co/datasets/soldni/jeopardy/commit/
|
364 |
]
|
365 |
},
|
366 |
-
"execution_count":
|
367 |
"metadata": {},
|
368 |
"output_type": "execute_result"
|
369 |
}
|
@@ -372,12 +391,15 @@
|
|
372 |
"all_questions = []\n",
|
373 |
"mosaicml_gauntlet = []\n",
|
374 |
"for row in old_questions:\n",
|
375 |
-
" key = '
|
376 |
" all_questions.append(row)\n",
|
377 |
" row['mosaicml_gauntlet'] = key in new_questions\n",
|
378 |
" if row['mosaicml_gauntlet']:\n",
|
379 |
" mosaicml_gauntlet.append(row)\n",
|
380 |
"\n",
|
|
|
|
|
|
|
381 |
"mosaicml_gauntlet_dataset = datasets.Dataset.from_list(mosaicml_gauntlet)\n",
|
382 |
"all_questions_dataset = datasets.Dataset.from_list(all_questions)\n",
|
383 |
"\n",
|
|
|
258 |
},
|
259 |
{
|
260 |
"cell_type": "code",
|
261 |
+
"execution_count": 17,
|
262 |
"metadata": {},
|
263 |
"outputs": [
|
264 |
{
|
265 |
"name": "stdout",
|
266 |
"output_type": "stream",
|
267 |
"text": [
|
268 |
+
"Missing: 0\n",
|
269 |
+
"Old questions: 2220\n",
|
270 |
+
"New questions: 2113\n"
|
271 |
]
|
272 |
}
|
273 |
],
|
|
|
277 |
"\n",
|
278 |
"missing = [{**v, 'key': k} for k, v in new_questions.items() if k not in old_questions_subset]\n",
|
279 |
"print(f\"Missing: {len(missing)}\")\n",
|
280 |
+
"print(f\"Old questions: {len(old_questions_subset)}\")\n",
|
281 |
+
"print(f\"New questions: {len(new_questions)}\")\n",
|
282 |
"for m in missing:\n",
|
283 |
" print(m['context'])\n",
|
284 |
" print(m['continuation'])\n",
|
|
|
295 |
},
|
296 |
{
|
297 |
"cell_type": "code",
|
298 |
+
"execution_count": 19,
|
299 |
"metadata": {},
|
300 |
"outputs": [
|
301 |
+
{
|
302 |
+
"name": "stdout",
|
303 |
+
"output_type": "stream",
|
304 |
+
"text": [
|
305 |
+
"Mosaicml Gauntlet: 2116\n",
|
306 |
+
"All questions: 216930\n"
|
307 |
+
]
|
308 |
+
},
|
309 |
{
|
310 |
"data": {
|
311 |
"application/vnd.jupyter.widget-view+json": {
|
312 |
+
"model_id": "b0db52ff66c84083984ce81539bb3129",
|
313 |
"version_major": 2,
|
314 |
"version_minor": 0
|
315 |
},
|
|
|
323 |
{
|
324 |
"data": {
|
325 |
"application/vnd.jupyter.widget-view+json": {
|
326 |
+
"model_id": "e07846cf593e47e78cc7c62d179c5baa",
|
327 |
"version_major": 2,
|
328 |
"version_minor": 0
|
329 |
},
|
330 |
"text/plain": [
|
331 |
+
"Creating parquet from Arrow format: 0%| | 0/3 [00:00<?, ?ba/s]"
|
332 |
]
|
333 |
},
|
334 |
"metadata": {},
|
335 |
"output_type": "display_data"
|
336 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
337 |
{
|
338 |
"data": {
|
339 |
"application/vnd.jupyter.widget-view+json": {
|
340 |
+
"model_id": "af4036c1ed13441c9a6cc1de5558dfa3",
|
341 |
"version_major": 2,
|
342 |
"version_minor": 0
|
343 |
},
|
|
|
351 |
{
|
352 |
"data": {
|
353 |
"application/vnd.jupyter.widget-view+json": {
|
354 |
+
"model_id": "e23eca2e0aac433baba52069b40c2e3a",
|
355 |
"version_major": 2,
|
356 |
"version_minor": 0
|
357 |
},
|
|
|
362 |
"metadata": {},
|
363 |
"output_type": "display_data"
|
364 |
},
|
365 |
+
{
|
366 |
+
"data": {
|
367 |
+
"application/vnd.jupyter.widget-view+json": {
|
368 |
+
"model_id": "7a6982d43029428c98258352ec90bf28",
|
369 |
+
"version_major": 2,
|
370 |
+
"version_minor": 0
|
371 |
+
},
|
372 |
+
"text/plain": [
|
373 |
+
"README.md: 0%| | 0.00/2.34k [00:00<?, ?B/s]"
|
374 |
+
]
|
375 |
+
},
|
376 |
+
"metadata": {},
|
377 |
+
"output_type": "display_data"
|
378 |
+
},
|
379 |
{
|
380 |
"data": {
|
381 |
"text/plain": [
|
382 |
+
"CommitInfo(commit_url='https://huggingface.co/datasets/soldni/jeopardy/commit/95da23525ad41487becfaf821be025ccefda6a34', commit_message='Upload dataset', commit_description='', oid='95da23525ad41487becfaf821be025ccefda6a34', pr_url=None, pr_revision=None, pr_num=None)"
|
383 |
]
|
384 |
},
|
385 |
+
"execution_count": 19,
|
386 |
"metadata": {},
|
387 |
"output_type": "execute_result"
|
388 |
}
|
|
|
391 |
"all_questions = []\n",
|
392 |
"mosaicml_gauntlet = []\n",
|
393 |
"for row in old_questions:\n",
|
394 |
+
" key = ''.join([ch for ch in f\"{row['context']} {row['continuation']}\" if ch not in punctuation and ch not in whitespace])\n",
|
395 |
" all_questions.append(row)\n",
|
396 |
" row['mosaicml_gauntlet'] = key in new_questions\n",
|
397 |
" if row['mosaicml_gauntlet']:\n",
|
398 |
" mosaicml_gauntlet.append(row)\n",
|
399 |
"\n",
|
400 |
+
"print(f\"Mosaicml Gauntlet: {len(mosaicml_gauntlet)}\")\n",
|
401 |
+
"print(f\"All questions: {len(all_questions)}\")\n",
|
402 |
+
"\n",
|
403 |
"mosaicml_gauntlet_dataset = datasets.Dataset.from_list(mosaicml_gauntlet)\n",
|
404 |
"all_questions_dataset = datasets.Dataset.from_list(all_questions)\n",
|
405 |
"\n",
|