Maurice Weber
commited on
Commit
•
167a6e2
1
Parent(s):
3654bf8
add snippet, fix citation
Browse files
README.md
CHANGED
@@ -77,6 +77,69 @@ done
|
|
77 |
A full set of scripts to recreate the dataset, including the quality signals, can be
|
78 |
found [here](https://github.com/togethercomputer/RedPajama-Data).
|
79 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
80 |
### Dataset Summary
|
81 |
|
82 |
RedPajama-V2 is an open dataset for training large laguage models and includes over 100B text documents. Out of these,
|
@@ -272,7 +335,7 @@ To cite RedPajama-V2, please use:
|
|
272 |
```
|
273 |
@software{together2023redpajama-v2,
|
274 |
author = {Together Computer},
|
275 |
-
title = {RedPajama
|
276 |
month = October,
|
277 |
year = 2023,
|
278 |
url = {https://github.com/togethercomputer/RedPajama-Data}
|
|
|
77 |
A full set of scripts to recreate the dataset, including the quality signals, can be
|
78 |
found [here](https://github.com/togethercomputer/RedPajama-Data).
|
79 |
|
80 |
+
### Applying Filtering Rules
|
81 |
+
You can use the quality signals to filter the raw RedPajama-V2 dataset for a given set of rules. For example, consider
|
82 |
+
the following set of rules used in Gopher:
|
83 |
+
|
84 |
+
```python
|
85 |
+
def gopher_rules_pass(sample) -> bool:
|
86 |
+
""" function returns True if the sample complies with Gopher rules """
|
87 |
+
signals = json.loads(sample["quality_signals"])
|
88 |
+
|
89 |
+
# rule 1: number of words between 50 and 10'000
|
90 |
+
word_count = signals["rps_doc_word_count"][0][2]
|
91 |
+
if word_count < 50 or word_count > 10_000:
|
92 |
+
return False
|
93 |
+
|
94 |
+
# rule 2: mean word length between 3 and 10
|
95 |
+
mean_word_length = signals["rps_doc_mean_word_length"][0][2]
|
96 |
+
if mean_word_length < 3 or mean_word_length > 10:
|
97 |
+
return False
|
98 |
+
|
99 |
+
# rule 2: symbol to word ratio below 0.1
|
100 |
+
symbol_word_ratio = signals["rps_doc_symbol_to_word_ratio"][0][2]
|
101 |
+
if symbol_word_ratio > 0.1:
|
102 |
+
return False
|
103 |
+
|
104 |
+
# rule 3: 90% of lines need to start without a bullet point
|
105 |
+
n_lines = signals["ccnet_nlines"][0][2]
|
106 |
+
n_lines_bulletpoint_start = sum(map(lambda ln: ln[2], signals["rps_lines_start_with_bulletpoint"]))
|
107 |
+
if n_lines_bulletpoint_start / n_lines > 0.9:
|
108 |
+
return False
|
109 |
+
|
110 |
+
# rule 4: the ratio between characters in the most frequent 2-gram and the total number
|
111 |
+
# of characters must be below 0.2
|
112 |
+
top_2_gram_frac = signals["rps_doc_frac_chars_top_2gram"][0][2]
|
113 |
+
if top_2_gram_frac > 0.2:
|
114 |
+
return False
|
115 |
+
|
116 |
+
# rule 5: ...
|
117 |
+
|
118 |
+
|
119 |
+
return True
|
120 |
+
```
|
121 |
+
|
122 |
+
Filtering the RedPajama-V2 dataset with this set of rules is then as easy as:
|
123 |
+
|
124 |
+
```python
|
125 |
+
ds_iterator = load_dataset(
|
126 |
+
"togethercomputer/RedPajama-Data-V2",
|
127 |
+
snapshots=["2023-14"],
|
128 |
+
languages=["en"],
|
129 |
+
name="default",
|
130 |
+
streaming=True
|
131 |
+
)
|
132 |
+
|
133 |
+
filtered_dataset = []
|
134 |
+
|
135 |
+
for sample in ds_iterator["train"]:
|
136 |
+
|
137 |
+
if not gopher_rules_pass(sample):
|
138 |
+
continue
|
139 |
+
|
140 |
+
filtered_dataset.append(sample)
|
141 |
+
```
|
142 |
+
|
143 |
### Dataset Summary
|
144 |
|
145 |
RedPajama-V2 is an open dataset for training large laguage models and includes over 100B text documents. Out of these,
|
|
|
335 |
```
|
336 |
@software{together2023redpajama-v2,
|
337 |
author = {Together Computer},
|
338 |
+
title = {RedPajama: an Open Dataset for Training Large Language Models},
|
339 |
month = October,
|
340 |
year = 2023,
|
341 |
url = {https://github.com/togethercomputer/RedPajama-Data}
|