Dataset Viewer
Auto-converted to Parquet Duplicate
run
stringclasses
1 value
step
int64
1
300
row
int64
0
98.3k
src
stringlengths
6
10
ctx_tail
stringlengths
35
246
true_next16
stringlengths
11
140
g
int64
0
7
text
stringlengths
3
161
reward
float64
0
0.81
cos
float64
-0.14
0.9
len
int64
3
18
lp_engine
float64
-141.86
-0.02
lp_train
float64
-142.05
-0.02
ids
stringlengths
17
134
mF_rl1
1
45,739
[1, 12971]
for peanuts). My constant tossing incorporated the salt evenly and when the peanuts became a light yellow with dark brown spots, I removed them from the heat into a medium
bowl. Along with the 3/4 cup peanuts, I assembled 1
0
* bowl. Again, I wanted these nuts to have
0.256147
0.506109
12
-28.329788
-28.38007
[22, 33116, 26, 26700, 24, 372, 12529, 2037, 56594, 328, 847, 200008]
mF_rl1
1
45,739
[1, 12971]
for peanuts). My constant tossing incorporated the salt evenly and when the peanuts became a light yellow with dark brown spots, I removed them from the heat into a medium
bowl. Along with the 3/4 cup peanuts, I assembled 1
1
* bowl sitting to the side of the stovetop.
0.254025
0.504009
14
-19.629305
-19.69397
[22, 33116, 29926, 328, 290, 4635, 323, 290, 384, 569, 296, 510, 26, 200008]
mF_rl1
1
45,739
[1, 12971]
for peanuts). My constant tossing incorporated the salt evenly and when the peanuts became a light yellow with dark brown spots, I removed them from the heat into a medium
bowl. Along with the 3/4 cup peanuts, I assembled 1
2
* bowl, add in the sliced almonds. - Then,
0.222541
0.471743
13
-31.526852
-31.512094
[22, 33116, 24, 1208, 310, 290, 94888, 168891, 26, 430, 6394, 24, 200008]
mF_rl1
1
45,739
[1, 12971]
for peanuts). My constant tossing incorporated the salt evenly and when the peanuts became a light yellow with dark brown spots, I removed them from the heat into a medium
bowl. Along with the 3/4 cup peanuts, I assembled 1
3
* bowl. - – Andrew
0.105832
0.325318
9
-26.362843
-26.618959
[22, 33116, 26, 220, 430, 220, 1071, 28304, 200008]
mF_rl1
1
45,739
[1, 12971]
for peanuts). My constant tossing incorporated the salt evenly and when the peanuts became a light yellow with dark brown spots, I removed them from the heat into a medium
bowl. Along with the 3/4 cup peanuts, I assembled 1
4
* soup pot, then add the tomato
0.14685
0.38321
9
-27.337402
-27.283905
[22, 34477, 2560, 24, 1740, 1208, 290, 66364, 200008]
mF_rl1
1
45,739
[1, 12971]
for peanuts). My constant tossing incorporated the salt evenly and when the peanuts became a light yellow with dark brown spots, I removed them from the heat into a medium
bowl. Along with the 3/4 cup peanuts, I assembled 1
5
* bowl. The bowl of melted butter can
0.212832
0.461337
10
-22.959122
-22.884245
[22, 33116, 26, 589, 33116, 323, 91218, 24032, 713, 200008]
mF_rl1
1
45,739
[1, 12971]
for peanuts). My constant tossing incorporated the salt evenly and when the peanuts became a light yellow with dark brown spots, I removed them from the heat into a medium
bowl. Along with the 3/4 cup peanuts, I assembled 1
6
* Pyrex measuring cup (for
0.156163
0.395175
8
-19.094372
-19.022194
[22, 11451, 88983, 22534, 8094, 330, 540, 200008]
mF_rl1
1
45,739
[1, 12971]
for peanuts). My constant tossing incorporated the salt evenly and when the peanuts became a light yellow with dark brown spots, I removed them from the heat into a medium
bowl. Along with the 3/4 cup peanuts, I assembled 1
7
* -sized bowl. Add sugar and salt to the mixture, and dump the
0.374965
0.612344
18
-31.701784
-31.669874
[22, 430, 95, 2158, 33116, 26, 3878, 18885, 341, 15787, 328, 290, 19413, 24, 341, 29866, 290, 200008]
mF_rl1
1
83,149
[2, 17613]
explicitly rejected by Protestantism, which demanded instead that a general Christian ethic permeate all aspects of an individual’s life.” While Catholicism surely has an ascetic
stream, the point is that the ancient streams of Christianity lived with these various tensions
0
* tradition, it also may not be as influential in our lived experience as
0.179626
0.423823
16
-36.958027
-37.040817
[22, 25017, 24, 536, 1437, 1719, 702, 446, 486, 71632, 310, 1753, 30290, 5726, 486, 200008]
mF_rl1
1
83,149
[2, 17613]
explicitly rejected by Protestantism, which demanded instead that a general Christian ethic permeate all aspects of an individual’s life.” While Catholicism surely has an ascetic
stream, the point is that the ancient streams of Christianity lived with these various tensions
1
* tradition, G.K. Chesterton highlights the relationship between asc
0.214952
0.46363
14
-27.220303
-27.092056
[22, 25017, 24, 477, 14277, 26, 104389, 62966, 38931, 290, 6632, 1867, 22892, 200008]
mF_rl1
1
83,149
[2, 17613]
explicitly rejected by Protestantism, which demanded instead that a general Christian ethic permeate all aspects of an individual’s life.” While Catholicism surely has an ascetic
stream, the point is that the ancient streams of Christianity lived with these various tensions
2
* vein though with respect to Eastern Spirituality (particularly Buddhism)
0.064697
0.254356
14
-48.99979
-49.162693
[22, 89420, 8174, 517, 4279, 328, 33471, 104300, 520, 330, 163798, 155971, 21, 200008]
mF_rl1
1
83,149
[2, 17613]
explicitly rejected by Protestantism, which demanded instead that a general Christian ethic permeate all aspects of an individual’s life.” While Catholicism surely has an ascetic
stream, the point is that the ancient streams of Christianity lived with these various tensions
3
* tradition, it can
0.151417
0.389124
6
-9.767137
-9.763398
[22, 25017, 24, 536, 713, 200008]
mF_rl1
1
83,149
[2, 17613]
explicitly rejected by Protestantism, which demanded instead that a general Christian ethic permeate all aspects of an individual’s life.” While Catholicism surely has an ascetic
stream, the point is that the ancient streams of Christianity lived with these various tensions
4
* tradition, this period of isolation that Fr. James Martin pronounces as
0.15672
0.395879
16
-52.842251
-52.566971
[22, 25017, 24, 544, 4336, 323, 49859, 511, 2717, 26, 14160, 16423, 17591, 65169, 486, 200008]
mF_rl1
1
83,149
[2, 17613]
explicitly rejected by Protestantism, which demanded instead that a general Christian ethic permeate all aspects of an individual’s life.” While Catholicism surely has an ascetic
stream, the point is that the ancient streams of Christianity lived with these various tensions
5
* tradition as well,
0.20683
0.454786
6
-7.567747
-7.541088
[22, 25017, 486, 2515, 24, 200008]
mF_rl1
1
83,149
[2, 17613]
explicitly rejected by Protestantism, which demanded instead that a general Christian ethic permeate all aspects of an individual’s life.” While Catholicism surely has an ascetic
stream, the point is that the ancient streams of Christianity lived with these various tensions
6
* tradition, so do modern
0.150836
0.388376
7
-13.436184
-13.359629
[22, 25017, 24, 1017, 641, 9888, 200008]
mF_rl1
1
83,149
[2, 17613]
explicitly rejected by Protestantism, which demanded instead that a general Christian ethic permeate all aspects of an individual’s life.” While Catholicism surely has an ascetic
stream, the point is that the ancient streams of Christianity lived with these various tensions
7
* tradition, one with much to offer our Protestant brothers and sisters
0.178554
0.422557
14
-31.848381
-31.977295
[22, 25017, 24, 1085, 517, 2639, 328, 5408, 1753, 146863, 55854, 341, 77862, 200008]
mF_rl1
1
75,539
[2, 10003]
and maximise our capacity to react. These will be existentially important in rapidly expanding coastal cities such as Buenos Aires and Mumbai, that are the most at risk to
exogenous shocks and some of the least equipped to subsequently deal with them. The World
0
* the worst ravages of flooding.
0.22019
0.469244
9
-19.834642
-19.86784
[22, 290, 29616, 56536, 1473, 323, 81864, 26, 200008]
mF_rl1
1
75,539
[2, 10003]
and maximise our capacity to react. These will be existentially important in rapidly expanding coastal cities such as Buenos Aires and Mumbai, that are the most at risk to
exogenous shocks and some of the least equipped to subsequently deal with them. The World
1
* the impacts of climate change, with sea-level rise
0.326181
0.571123
12
-17.128078
-17.096272
[22, 290, 33256, 323, 17053, 3332, 24, 517, 10894, 15636, 16980, 200008]
mF_rl1
1
75,539
[2, 10003]
and maximise our capacity to react. These will be existentially important in rapidly expanding coastal cities such as Buenos Aires and Mumbai, that are the most at risk to
exogenous shocks and some of the least equipped to subsequently deal with them. The World
2
* climate change, such
0.244031
0.493995
6
-9.794904
-9.710224
[22, 17053, 3332, 24, 2203, 200008]
mF_rl1
1
75,539
[2, 10003]
and maximise our capacity to react. These will be existentially important in rapidly expanding coastal cities such as Buenos Aires and Mumbai, that are the most at risk to
exogenous shocks and some of the least equipped to subsequently deal with them. The World
3
* climate shifts. In this sense, public art is all
0.178893
0.422958
13
-35.932018
-35.964493
[22, 17053, 40955, 26, 608, 544, 10203, 24, 854, 2265, 373, 806, 200008]
mF_rl1
1
75,539
[2, 10003]
and maximise our capacity to react. These will be existentially important in rapidly expanding coastal cities such as Buenos Aires and Mumbai, that are the most at risk to
exogenous shocks and some of the least equipped to subsequently deal with them. The World
4
* our climate changing. But a comparable amount of capital must be deployed to make the
0.188834
0.43455
18
-56.368263
-56.570683
[22, 1753, 17053, 16005, 26, 3696, 262, 36734, 3882, 323, 7963, 3066, 446, 46813, 328, 2170, 290, 200008]
mF_rl1
1
75,539
[2, 10003]
and maximise our capacity to react. These will be existentially important in rapidly expanding coastal cities such as Buenos Aires and Mumbai, that are the most at risk to
exogenous shocks and some of the least equipped to subsequently deal with them. The World
5
* future climate risks, historical patterns need
0.108775
0.329811
9
-37.223671
-37.175575
[22, 6482, 17053, 23981, 24, 18602, 15268, 1657, 200008]
mF_rl1
1
75,539
[2, 10003]
and maximise our capacity to react. These will be existentially important in rapidly expanding coastal cities such as Buenos Aires and Mumbai, that are the most at risk to
exogenous shocks and some of the least equipped to subsequently deal with them. The World
6
* climate change. Nor are they threatened by it. They are biggest perpetrators of
0.206538
0.454464
18
-47.230976
-47.00518
[22, 17053, 3332, 26, 14560, 583, 1451, 72934, 618, 536, 26, 4647, 583, 21989, 42295, 113338, 323, 200008]
mF_rl1
1
75,539
[2, 10003]
and maximise our capacity to react. These will be existentially important in rapidly expanding coastal cities such as Buenos Aires and Mumbai, that are the most at risk to
exogenous shocks and some of the least equipped to subsequently deal with them. The World
7
* climate change - such as rising sea levels, poor weather
0.239956
0.489853
13
-29.55982
-29.585636
[22, 17053, 3332, 430, 2203, 486, 32448, 10894, 8248, 24, 15720, 15040, 200008]
mF_rl1
1
74,952
[2, 9416]
the world we agreed on 40 designers in total. The British designers were easy as I knew of many, the hardest I found were the Italian, which really
surprised me as Italy is well know for Knitwear. Just as I was
0
* only supplied Palais Gui
0.037384
0.19335
6
-39.674721
-39.889488
[22, 1809, 29812, 188921, 85173, 200008]
mF_rl1
1
74,952
[2, 9416]
the world we agreed on 40 designers in total. The British designers were easy as I knew of many, the hardest I found were the Italian, which really
surprised me as Italy is well know for Knitwear. Just as I was
1
* only produce a handful of geniuses.
0.04579
0.213986
11
-25.510572
-25.67185
[22, 1809, 10021, 262, 70449, 323, 3698, 85, 10123, 26, 200008]
mF_rl1
1
74,952
[2, 9416]
the world we agreed on 40 designers in total. The British designers were easy as I knew of many, the hardest I found were the Italian, which really
surprised me as Italy is well know for Knitwear. Just as I was
2
* aren’t recognised for their design merit? For instance in Paris
0.045748
0.213887
14
-49.909107
-50.056202
[22, 19311, 2705, 84042, 393, 1357, 4174, 62381, 43, 1972, 5079, 310, 13796, 200008]
mF_rl1
1
74,952
[2, 9416]
the world we agreed on 40 designers in total. The British designers were easy as I knew of many, the hardest I found were the Italian, which really
surprised me as Italy is well know for Knitwear. Just as I was
3
* surprised me personnally as I thought Spain design
0.139891
0.37402
12
-37.274803
-37.467136
[22, 40858, 768, 2084, 90, 3186, 486, 372, 7956, 31154, 4174, 200008]
mF_rl1
1
74,952
[2, 9416]
the world we agreed on 40 designers in total. The British designers were easy as I knew of many, the hardest I found were the Italian, which really
surprised me as Italy is well know for Knitwear. Just as I was
4
* struggled as most of their pieces were still in the early stages. My good
0.029024
0.170365
17
-45.862518
-45.876118
[22, 83513, 486, 2167, 323, 1357, 18326, 1610, 4296, 310, 290, 8201, 26520, 26, 4902, 3049, 200008]
mF_rl1
1
74,952
[2, 9416]
the world we agreed on 40 designers in total. The British designers were easy as I knew of many, the hardest I found were the Italian, which really
surprised me as Italy is well know for Knitwear. Just as I was
5
* were just gelled so well, put together in such a nicely executed way
0.046583
0.21583
17
-63.583931
-63.698265
[22, 1610, 1910, 10006, 1146, 1017, 2515, 24, 4276, 6473, 310, 2203, 262, 70603, 26267, 2845, 200008]
mF_rl1
1
74,952
[2, 9416]
the world we agreed on 40 designers in total. The British designers were easy as I knew of many, the hardest I found were the Italian, which really
surprised me as Italy is well know for Knitwear. Just as I was
6
* surprised me; for such a young movement with little commercial backing (
0.150108
0.387438
15
-41.902939
-41.927696
[22, 40858, 768, 39, 393, 2203, 262, 10844, 14510, 517, 6060, 13732, 69643, 330, 200008]
mF_rl1
1
74,952
[2, 9416]
the world we agreed on 40 designers in total. The British designers were easy as I knew of many, the hardest I found were the Italian, which really
surprised me as Italy is well know for Knitwear. Just as I was
7
* surprised me except one did and that was the Mexican School, as I’d not
0.102709
0.320483
18
-50.335007
-50.288612
[22, 40858, 768, 9171, 1085, 3129, 341, 511, 812, 290, 39800, 8839, 24, 486, 372, 12450, 702, 200008]
mF_rl1
1
50,667
[1, 17899]
7.50 4Videosoft Mac iPod Touch Manager for ePub is Provides an excellent platform for iPod Touch users. It possesses the ability of
transfering ePub to iPod Touch and ePub from iPod Touch software
0
* backup ability to transferring iPhone/iPod touch to Windows PC or
0.24378
0.493741
15
-42.599018
-42.496948
[22, 28628, 10320, 328, 89563, 597, 8121, 24520, 22980, 11675, 328, 6706, 11294, 537, 200008]
mF_rl1
1
50,667
[1, 17899]
7.50 4Videosoft Mac iPod Touch Manager for ePub is Provides an excellent platform for iPod Touch users. It possesses the ability of
transfering ePub to iPod Touch and ePub from iPod Touch software
1
* copying audio books, video music, podcast, ringtone, photo, etc.
0.254559
0.504538
18
-42.272877
-42.228508
[22, 70988, 13931, 12504, 24, 6061, 9184, 24, 37628, 24, 14012, 150026, 24, 14460, 24, 6541, 26, 200008]
mF_rl1
1
50,667
[1, 17899]
7.50 4Videosoft Mac iPod Touch Manager for ePub is Provides an excellent platform for iPod Touch users. It possesses the ability of
transfering ePub to iPod Touch and ePub from iPod Touch software
2
* making iTunes backup, iPod data file transfer, managing iOS file with
0.343948
0.586471
18
-42.725143
-42.712303
[22, 6088, 597, 50205, 28628, 24, 597, 22980, 1049, 2017, 7088, 24, 30732, 597, 3154, 2017, 517, 200008]
mF_rl1
1
50,667
[1, 17899]
7.50 4Videosoft Mac iPod Touch Manager for ePub is Provides an excellent platform for iPod Touch users. It possesses the ability of
transfering ePub to iPod Touch and ePub from iPod Touch software
3
* transfer music from iPod to
0.208644
0.456776
8
-11.501713
-11.547698
[22, 7088, 9184, 610, 597, 22980, 328, 200008]
mF_rl1
1
50,667
[1, 17899]
7.50 4Videosoft Mac iPod Touch Manager for ePub is Provides an excellent platform for iPod Touch users. It possesses the ability of
transfering ePub to iPod Touch and ePub from iPod Touch software
4
* syncing music and photos between
0.209668
0.457895
8
-14.925838
-14.998189
[22, 27290, 300, 9184, 341, 17623, 1867, 200008]
mF_rl1
1
50,667
[1, 17899]
7.50 4Videosoft Mac iPod Touch Manager for ePub is Provides an excellent platform for iPod Touch users. It possesses the ability of
transfering ePub to iPod Touch and ePub from iPod Touch software
5
* helping users
0.035924
0.189537
4
-15.052219
-15.001092
[22, 23242, 6755, 200008]
mF_rl1
1
50,667
[1, 17899]
7.50 4Videosoft Mac iPod Touch Manager for ePub is Provides an excellent platform for iPod Touch users. It possesses the ability of
transfering ePub to iPod Touch and ePub from iPod Touch software
6
* downloading files from iTunes, photos or videos from
0.137135
0.370317
12
-30.090046
-30.072372
[22, 73530, 5743, 610, 597, 50205, 24, 17623, 537, 20755, 610, 200008]
mF_rl1
1
50,667
[1, 17899]
7.50 4Videosoft Mac iPod Touch Manager for ePub is Provides an excellent platform for iPod Touch users. It possesses the ability of
transfering ePub to iPod Touch and ePub from iPod Touch software
7
* transferring files between iOS and
0.24844
0.498438
8
-11.895303
-11.86043
[22, 89563, 5743, 1867, 597, 3154, 341, 200008]
mF_rl1
1
8,831
[0, 8831]
. Current Status: The relocation completed in December 2017. Mainline Signalling: Growth Deal contribution: £3.2m The scheme provides local
investment in rail infrastructure to allow half hourly mainline services. The alignment of signalling
0
* match funding of £2.5m for this project (Hastings
0.095217
0.308572
16
-35.282742
-35.21207
[22, 4321, 25094, 323, 15767, 30, 26, 33, 89, 393, 544, 3362, 330, 169499, 1104, 200008]
mF_rl1
1
8,831
[0, 8831]
. Current Status: The relocation completed in December 2017. Mainline Signalling: Growth Deal contribution: £3.2m The scheme provides local
investment in rail infrastructure to allow half hourly mainline services. The alignment of signalling
1
* contributions towards the costs of the signalling and control
0.113412
0.336767
11
-19.552158
-19.569357
[22, 24929, 12315, 290, 7662, 323, 290, 169911, 341, 3266, 200008]
mF_rl1
1
8,831
[0, 8831]
. Current Status: The relocation completed in December 2017. Mainline Signalling: Growth Deal contribution: £3.2m The scheme provides local
investment in rail infrastructure to allow half hourly mainline services. The alignment of signalling
2
* contribution to real-time control of the railway by demonstrating the benefits
0.0487
0.220681
14
-33.395321
-33.431587
[22, 20070, 328, 1776, 10457, 3266, 323, 290, 69704, 618, 67468, 290, 12093, 200008]
mF_rl1
1
8,831
[0, 8831]
. Current Status: The relocation completed in December 2017. Mainline Signalling: Growth Deal contribution: £3.2m The scheme provides local
investment in rail infrastructure to allow half hourly mainline services. The alignment of signalling
3
* signalling between East Kent and Ashford, enabling local traffic to enter
0.105976
0.32554
15
-39.534004
-39.187454
[22, 169911, 1867, 13902, 32834, 341, 26223, 6860, 24, 41104, 2743, 16639, 328, 6668, 200008]
mF_rl1
1
8,831
[0, 8831]
. Current Status: The relocation completed in December 2017. Mainline Signalling: Growth Deal contribution: £3.2m The scheme provides local
investment in rail infrastructure to allow half hourly mainline services. The alignment of signalling
4
* contribution to start capacity and
0.07777
0.278873
7
-25.758265
-25.611736
[22, 20070, 328, 1965, 11610, 341, 200008]
mF_rl1
1
8,831
[0, 8831]
. Current Status: The relocation completed in December 2017. Mainline Signalling: Growth Deal contribution: £3.2m The scheme provides local
investment in rail infrastructure to allow half hourly mainline services. The alignment of signalling
5
* rail infrastructure enhancements, particularly on the Marsh
0.097652
0.312493
10
-21.139318
-21.031952
[22, 26261, 25073, 108528, 24, 14445, 436, 290, 84854, 200008]
mF_rl1
1
8,831
[0, 8831]
. Current Status: The relocation completed in December 2017. Mainline Signalling: Growth Deal contribution: £3.2m The scheme provides local
investment in rail infrastructure to allow half hourly mainline services. The alignment of signalling
6
* enhancements to the signalling system from 2014 to provide additional capacity on
0.081802
0.28601
16
-27.76712
-27.884348
[22, 108528, 328, 290, 169911, 1829, 610, 220, 703, 32, 328, 4706, 7445, 11610, 436, 200008]
mF_rl1
1
8,831
[0, 8831]
. Current Status: The relocation completed in December 2017. Mainline Signalling: Growth Deal contribution: £3.2m The scheme provides local
investment in rail infrastructure to allow half hourly mainline services. The alignment of signalling
7
* signalling infrastructure. In
0.072683
0.269597
6
-16.12326
-16.163801
[22, 169911, 25073, 26, 608, 200008]
mF_rl1
1
61,704
[1, 28936]
when I want something cheesy. It should not be confused with brewer's yeast which tastes bitter, or the active yeast used for leavening. It is
inexpensive and readily available packaged or in bulk at natural food stores. Tofu-
0
* strongest smelling yeast you’ll find and is
0.069141
0.262947
11
-36.539444
-36.435341
[22, 64876, 2065, 5869, 73318, 650, 8874, 1851, 341, 373, 200008]
mF_rl1
1
61,704
[1, 28936]
when I want something cheesy. It should not be confused with brewer's yeast which tastes bitter, or the active yeast used for leavening. It is
inexpensive and readily available packaged or in bulk at natural food stores. Tofu-
1
* a dry yeast, so it keeps for a very long
0.235844
0.485638
13
-22.368111
-22.359324
[22, 262, 13289, 73318, 24, 1017, 536, 30095, 393, 262, 3044, 1936, 200008]
mF_rl1
1
61,704
[1, 28936]
when I want something cheesy. It should not be confused with brewer's yeast which tastes bitter, or the active yeast used for leavening. It is
inexpensive and readily available packaged or in bulk at natural food stores. Tofu-
2
* sold in health food stores but can be found in certain grocery
0.146713
0.383031
14
-21.895885
-21.892103
[22, 11464, 310, 4500, 6030, 21669, 1108, 713, 446, 3030, 310, 5598, 59880, 200008]
mF_rl1
1
61,704
[1, 28936]
when I want something cheesy. It should not be confused with brewer's yeast which tastes bitter, or the active yeast used for leavening. It is
inexpensive and readily available packaged or in bulk at natural food stores. Tofu-
3
* also available in supermarkets (106 + Uprising
0.216883
0.465707
11
-53.511627
-53.753143
[22, 1437, 4052, 310, 188627, 330, 10840, 584, 574, 81102, 200008]
mF_rl1
1
61,704
[1, 28936]
when I want something cheesy. It should not be confused with brewer's yeast which tastes bitter, or the active yeast used for leavening. It is
inexpensive and readily available packaged or in bulk at natural food stores. Tofu-
4
* usually sprinkled on flax or hemp seeds. - Have special
0.200267
0.447512
15
-46.979961
-46.975334
[22, 9441, 140182, 1146, 436, 1432, 842, 537, 125854, 33583, 26, 430, 23880, 5170, 200008]
mF_rl1
1
61,704
[1, 28936]
when I want something cheesy. It should not be confused with brewer's yeast which tastes bitter, or the active yeast used for leavening. It is
inexpensive and readily available packaged or in bulk at natural food stores. Tofu-
5
* also known as dried yeast as it is dead and
0.345378
0.587689
12
-29.328772
-29.439026
[22, 1437, 5711, 486, 54570, 73318, 486, 536, 373, 13655, 341, 200008]
mF_rl1
1
61,704
[1, 28936]
when I want something cheesy. It should not be confused with brewer's yeast which tastes bitter, or the active yeast used for leavening. It is
inexpensive and readily available packaged or in bulk at natural food stores. Tofu-
6
* found in bulk bins at health
0.253928
0.503913
8
-12.626077
-12.704166
[22, 3030, 310, 28458, 49605, 552, 4500, 200008]
mF_rl1
1
61,704
[1, 28936]
when I want something cheesy. It should not be confused with brewer's yeast which tastes bitter, or the active yeast used for leavening. It is
inexpensive and readily available packaged or in bulk at natural food stores. Tofu-
7
* available in most health food stores or online
0.355796
0.596487
10
-10.697979
-10.722706
[22, 4052, 310, 2167, 4500, 6030, 21669, 537, 5122, 200008]
mF_rl1
1
4,445
[0, 4445]
and long form, theatrical release, commercial and music video categories will receive a trophy and three days of all-expenses-paid training with Kevin Shaw, the company's
director of training, at da Vinci's facilities in Coral Springs, Fla. "This
0
* director of imaging and professional services. As part of the demonstration, Ke
0.234906
0.484671
16
-41.65712
-41.612442
[22, 15983, 323, 24880, 341, 11989, 5791, 26, 2188, 1030, 323, 290, 60484, 24, 8365, 200008]
mF_rl1
1
4,445
[0, 4445]
and long form, theatrical release, commercial and music video categories will receive a trophy and three days of all-expenses-paid training with Kevin Shaw, the company's
director of training, at da Vinci's facilities in Coral Springs, Fla. "This
1
* training director.
0.160739
0.400923
5
-9.716906
-9.754554
[22, 7962, 15983, 26, 200008]
mF_rl1
1
4,445
[0, 4445]
and long form, theatrical release, commercial and music video categories will receive a trophy and three days of all-expenses-paid training with Kevin Shaw, the company's
director of training, at da Vinci's facilities in Coral Springs, Fla. "This
2
* head of color services.
0.223101
0.472336
7
-13.27225
-13.133759
[22, 4234, 323, 2911, 5791, 26, 200008]
mF_rl1
1
4,445
[0, 4445]
and long form, theatrical release, commercial and music video categories will receive a trophy and three days of all-expenses-paid training with Kevin Shaw, the company's
director of training, at da Vinci's facilities in Coral Springs, Fla. "This
3
* workflow specialist. DaVinci Resolve 9
0.134115
0.366217
11
-16.684456
-16.697931
[22, 42214, 41703, 26, 10321, 66, 32180, 103554, 220, 37, 200008]
mF_rl1
1
4,445
[0, 4445]
and long form, theatrical release, commercial and music video categories will receive a trophy and three days of all-expenses-paid training with Kevin Shaw, the company's
director of training, at da Vinci's facilities in Coral Springs, Fla. "This
4
* senior color consultant.
0.23685
0.486672
6
-9.206608
-9.216524
[22, 24733, 2911, 66545, 26, 200008]
mF_rl1
1
4,445
[0, 4445]
and long form, theatrical release, commercial and music video categories will receive a trophy and three days of all-expenses-paid training with Kevin Shaw, the company's
director of training, at da Vinci's facilities in Coral Springs, Fla. "This
5
* director of training and director of marketing
0.14414
0.379657
9
-14.194848
-14.166018
[22, 15983, 323, 7962, 341, 15983, 323, 13335, 200008]
mF_rl1
1
4,445
[0, 4445]
and long form, theatrical release, commercial and music video categories will receive a trophy and three days of all-expenses-paid training with Kevin Shaw, the company's
director of training, at da Vinci's facilities in Coral Springs, Fla. "This
6
* director of product line management. FilmLight Updates Baselight, Daylight
0.267745
0.517441
15
-31.640228
-31.691696
[22, 15983, 323, 2543, 1978, 7768, 26, 25621, 18716, 54180, 91729, 594, 24, 161414, 200008]
mF_rl1
1
4,445
[0, 4445]
and long form, theatrical release, commercial and music video categories will receive a trophy and three days of all-expenses-paid training with Kevin Shaw, the company's
director of training, at da Vinci's facilities in Coral Springs, Fla. "This
7
* director of color. That experience includes "Mission
0.081576
0.285615
11
-33.573635
-33.47699
[22, 15983, 323, 2911, 26, 6390, 5726, 8107, 392, 86786, 200008]
mF_rl1
1
73,142
[2, 7606]
and Sunday. 6 Leith Festival Gala Day Leith Links, Edinburgh, Saturday An annual community-centric, carnival-themed mix of stalls, live performances,
fairground rides, and food and drink. Kicking off at 11am
0
* face-painting and activities for kids and adults
0.313842
0.560216
12
-19.826426
-19.818466
[22, 6518, 2656, 4701, 300, 341, 10540, 393, 16767, 341, 20464, 200008]
mF_rl1
1
73,142
[2, 7606]
and Sunday. 6 Leith Festival Gala Day Leith Links, Edinburgh, Saturday An annual community-centric, carnival-themed mix of stalls, live performances,
fairground rides, and food and drink. Kicking off at 11am
1
* rides and more. Before
0.26986
0.51948
7
-18.714222
-18.777109
[22, 69249, 341, 1211, 26, 22135, 200008]
mF_rl1
1
73,142
[2, 7606]
and Sunday. 6 Leith Festival Gala Day Leith Links, Edinburgh, Saturday An annual community-centric, carnival-themed mix of stalls, live performances,
fairground rides, and food and drink. Kicking off at 11am
2
* food and craft stalls, and more. It’s a very popular festival and
0.385823
0.621147
17
-33.981503
-34.031296
[22, 6030, 341, 23877, 173957, 24, 341, 1211, 26, 1573, 1082, 262, 3044, 7559, 33550, 341, 200008]
mF_rl1
1
73,142
[2, 7606]
and Sunday. 6 Leith Festival Gala Day Leith Links, Edinburgh, Saturday An annual community-centric, carnival-themed mix of stalls, live performances,
fairground rides, and food and drink. Kicking off at 11am
3
* children's activities and more – and what better way to enjoy fun and family-friendly events
0.364931
0.604095
18
-46.422112
-46.497314
[22, 78801, 10540, 341, 1211, 1071, 341, 1780, 4948, 2845, 328, 9737, 4352, 341, 5923, 31075, 7074, 200008]
mF_rl1
1
73,142
[2, 7606]
and Sunday. 6 Leith Festival Gala Day Leith Links, Edinburgh, Saturday An annual community-centric, carnival-themed mix of stalls, live performances,
fairground rides, and food and drink. Kicking off at 11am
4
* dance and music.
0.12132
0.34831
6
-10.940102
-11.023247
[22, 31473, 341, 9184, 26, 200008]
mF_rl1
1
73,142
[2, 7606]
and Sunday. 6 Leith Festival Gala Day Leith Links, Edinburgh, Saturday An annual community-centric, carnival-themed mix of stalls, live performances,
fairground rides, and food and drink. Kicking off at 11am
5
* and entertainment. “I
0.008942
0.094564
7
-18.002527
-18.04043
[22, 341, 36829, 26, 1092, 53, 200008]
mF_rl1
1
73,142
[2, 7606]
and Sunday. 6 Leith Festival Gala Day Leith Links, Edinburgh, Saturday An annual community-centric, carnival-themed mix of stalls, live performances,
fairground rides, and food and drink. Kicking off at 11am
6
* and funfest including facepainting, children's
0.069463
0.263559
11
-38.320183
-38.407646
[22, 341, 4352, 48145, 4414, 6518, 50832, 300, 24, 78801, 200008]
mF_rl1
1
73,142
[2, 7606]
and Sunday. 6 Leith Festival Gala Day Leith Links, Edinburgh, Saturday An annual community-centric, carnival-themed mix of stalls, live performances,
fairground rides, and food and drink. Kicking off at 11am
7
* food and workshops. Acts include Edinburgh Zoo
0.380247
0.616642
10
-29.264519
-29.264999
[22, 6030, 341, 63020, 26, 101934, 4440, 74333, 78516, 200008]
mF_rl1
1
25,612
[0, 25612]
GitHub online platforms and are expected to teach 10 highly demanded jobs. The training is available to workers across the globe and it is geared towards teaching digital skills
. The pandemic, having created a more technology-dependent face of work, and Microsoft
0
* to meet the demands of an uncertain
0.054833
0.234164
9
-19.039434
-19.021519
[22, 328, 7006, 290, 37389, 323, 315, 55066, 200008]
mF_rl1
1
25,612
[0, 25612]
GitHub online platforms and are expected to teach 10 highly demanded jobs. The training is available to workers across the globe and it is geared towards teaching digital skills
. The pandemic, having created a more technology-dependent face of work, and Microsoft
1
* . “We aim to expand access and opportunities through our upskilling
0.0572
0.239165
16
-39.276573
-39.212234
[22, 596, 1092, 2991, 13658, 328, 11818, 4089, 341, 17010, 2185, 1753, 1020, 2076, 10920, 200008]
mF_rl1
1
25,612
[0, 25612]
GitHub online platforms and are expected to teach 10 highly demanded jobs. The training is available to workers across the globe and it is geared towards teaching digital skills
. The pandemic, having created a more technology-dependent face of work, and Microsoft
2
* , including for beginners. The learners in
0.071774
0.267906
10
-32.139641
-32.172966
[22, 651, 4414, 393, 87210, 26, 589, 62186, 310, 200008]
mF_rl1
1
25,612
[0, 25612]
GitHub online platforms and are expected to teach 10 highly demanded jobs. The training is available to workers across the globe and it is geared towards teaching digital skills
. The pandemic, having created a more technology-dependent face of work, and Microsoft
3
* including Word, Excel,
0.053386
0.231053
7
-10.996592
-11.118439
[22, 4414, 13509, 24, 17392, 24, 200008]
mF_rl1
1
25,612
[0, 25612]
GitHub online platforms and are expected to teach 10 highly demanded jobs. The training is available to workers across the globe and it is geared towards teaching digital skills
. The pandemic, having created a more technology-dependent face of work, and Microsoft
4
* that can improve employment outcomes amid global
0.121638
0.348766
9
-27.473059
-27.489742
[22, 511, 713, 9264, 26704, 18909, 56644, 5420, 200008]
mF_rl1
1
25,612
[0, 25612]
GitHub online platforms and are expected to teach 10 highly demanded jobs. The training is available to workers across the globe and it is geared towards teaching digital skills
. The pandemic, having created a more technology-dependent face of work, and Microsoft
5
* for their workers. They have also organized conferences, hosted networking events
0.058753
0.24239
15
-47.257256
-47.36496
[22, 393, 1357, 15927, 26, 4647, 847, 1437, 31834, 85018, 24, 45738, 52552, 7074, 200008]
mF_rl1
1
25,612
[0, 25612]
GitHub online platforms and are expected to teach 10 highly demanded jobs. The training is available to workers across the globe and it is geared towards teaching digital skills
. The pandemic, having created a more technology-dependent face of work, and Microsoft
6
* . The first goal is to get workers to sign up for
0.049091
0.221565
14
-31.762667
-31.644468
[22, 596, 589, 1660, 10848, 373, 328, 872, 15927, 328, 1781, 1020, 393, 200008]
mF_rl1
1
25,612
[0, 25612]
GitHub online platforms and are expected to teach 10 highly demanded jobs. The training is available to workers across the globe and it is geared towards teaching digital skills
. The pandemic, having created a more technology-dependent face of work, and Microsoft
7
* and “re-skill” the unemployed. “As
0.076241
0.276118
13
-29.374416
-29.514431
[22, 341, 1092, 273, 1396, 31342, 838, 290, 120907, 26, 1092, 2922, 200008]
mF_rl1
1
7,929
[0, 7929]
challenges will be disqualified . Winner: Lily 1 That Colorful Lily! Challenge finished on Tuesday, 19 Jun 2012 Second challenge in the
Springtime Series. The first was "That Colorful Daisy" and this one
0
* “The Lily is in Full
0.04114
0.20283
8
-19.152851
-19.159266
[22, 1092, 954, 98624, 373, 310, 13751, 200008]
mF_rl1
1
7,929
[0, 7929]
challenges will be disqualified . Winner: Lily 1 That Colorful Lily! Challenge finished on Tuesday, 19 Jun 2012 Second challenge in the
Springtime Series. The first was "That Colorful Daisy" and this one
1
* Vergissmeinnicht growing in the garden contest
0.113609
0.33706
12
-39.329449
-39.119621
[22, 24725, 1096, 1099, 5317, 1945, 16002, 310, 290, 24619, 28362, 200008]
mF_rl1
1
7,929
[0, 7929]
challenges will be disqualified . Winner: Lily 1 That Colorful Lily! Challenge finished on Tuesday, 19 Jun 2012 Second challenge in the
Springtime Series. The first was "That Colorful Daisy" and this one
2
* Lily Universe and 20th Annual Vince and Joy Wiggins Exotic Plant
0.048466
0.22015
17
-65.615067
-65.773293
[22, 98624, 52465, 341, 220, 451, 418, 26425, 164648, 341, 52980, 484, 100179, 1559, 13293, 26362, 200008]
mF_rl1
1
7,929
[0, 7929]
challenges will be disqualified . Winner: Lily 1 That Colorful Lily! Challenge finished on Tuesday, 19 Jun 2012 Second challenge in the
Springtime Series. The first was "That Colorful Daisy" and this one
3
* "Bloom" series by Marigold Owls. The
0.108158
0.328874
14
-42.777039
-42.93335
[22, 392, 140548, 14, 5758, 618, 2921, 346, 1386, 75147, 5805, 26, 589, 200008]
mF_rl1
1
7,929
[0, 7929]
challenges will be disqualified . Winner: Lily 1 That Colorful Lily! Challenge finished on Tuesday, 19 Jun 2012 Second challenge in the
Springtime Series. The first was "That Colorful Daisy" and this one
4
* Lily Flower Series...like this:. I long
0.142274
0.377193
11
-45.336731
-45.235588
[22, 98624, 91576, 15436, 1043, 12506, 544, 32979, 372, 1936, 200008]
mF_rl1
1
7,929
[0, 7929]
challenges will be disqualified . Winner: Lily 1 That Colorful Lily! Challenge finished on Tuesday, 19 Jun 2012 Second challenge in the
Springtime Series. The first was "That Colorful Daisy" and this one
5
* series. Let's grow beautiful flowers once again!! I love
0.079465
0.281896
13
-43.410408
-43.377983
[22, 5758, 26, 29017, 4386, 14442, 35735, 6837, 3357, 4959, 372, 7123, 200008]
mF_rl1
1
7,929
[0, 7929]
challenges will be disqualified . Winner: Lily 1 That Colorful Lily! Challenge finished on Tuesday, 19 Jun 2012 Second challenge in the
Springtime Series. The first was "That Colorful Daisy" and this one
6
* "Single Flower" Dedicated sections:1.
0.076217
0.276074
11
-46.683838
-46.792984
[22, 392, 16738, 91576, 14, 190995, 17649, 38, 29, 26, 200008]
mF_rl1
1
7,929
[0, 7929]
challenges will be disqualified . Winner: Lily 1 That Colorful Lily! Challenge finished on Tuesday, 19 Jun 2012 Second challenge in the
Springtime Series. The first was "That Colorful Daisy" and this one
7
* Perennial flower series: Daylilies (
0.078843
0.28079
11
-24.21879
-24.116697
[22, 3594, 68245, 39417, 5758, 38, 9947, 163112, 601, 330, 200008]
mF_rl1
1
56,578
[1, 23810]
| |Car||she has a luxurious car.| |House||Biannca Raines has been known to live her own house.| |School||no verified
information found about school.| |College||No college found in our database.| |
0
* school records on their education, only found in High school
0.038883
0.197188
13
-42.332859
-42.597183
[22, 6408, 13308, 436, 1357, 10909, 24, 1809, 3030, 310, 8051, 6408, 200008]
mF_rl1
1
56,578
[1, 23810]
| |Car||she has a luxurious car.| |House||Biannca Raines has been known to live her own house.| |School||no verified
information found about school.| |College||No college found in our database.| |
1
* report available. -
0.087823
0.296349
6
-12.329369
-12.344066
[22, 3602, 4052, 26, 430, 200008]
mF_rl1
1
56,578
[1, 23810]
| |Car||she has a luxurious car.| |House||Biannca Raines has been known to live her own house.| |School||no verified
information found about school.| |College||No college found in our database.| |
2
* information yet. There is little explanation of what Sarah Paulson’s character does in
0.013195
0.114872
18
-44.826279
-44.859318
[22, 2688, 7441, 26, 3741, 373, 6060, 20966, 323, 1780, 37802, 12157, 1389, 1082, 4723, 2406, 310, 200008]
mF_rl1
1
56,578
[1, 23810]
| |Car||she has a luxurious car.| |House||Biannca Raines has been known to live her own house.| |School||no verified
information found about school.| |College||No college found in our database.| |
3
* sources available. Joined SAFTA
0.035941
0.189581
9
-35.848625
-35.801308
[22, 11730, 4052, 26, 12338, 1903, 20022, 110073, 200008]
mF_rl1
1
56,578
[1, 23810]
| |Car||she has a luxurious car.| |House||Biannca Raines has been known to live her own house.| |School||no verified
information found about school.| |College||No college found in our database.| |
4
* details. 2) Brittney is
0.034171
0.184854
10
-28.630878
-28.680384
[22, 7414, 26, 220, 30, 21, 122176, 5830, 373, 200008]
mF_rl1
1
56,578
[1, 23810]
| |Car||she has a luxurious car.| |House||Biannca Raines has been known to live her own house.| |School||no verified
information found about school.| |College||No college found in our database.| |
5
* data about his studies 4.What is School Beauty’s
0.052594
0.229333
13
-41.743221
-41.927273
[22, 1049, 1509, 1745, 9198, 220, 32, 85330, 373, 8839, 70472, 1082, 200008]
mF_rl1
1
56,578
[1, 23810]
| |Car||she has a luxurious car.| |House||Biannca Raines has been known to live her own house.| |School||no verified
information found about school.| |College||No college found in our database.| |
6
* information on that..
0.046675
0.216044
6
-14.752356
-14.710615
[22, 2688, 436, 511, 396, 200008]
mF_rl1
1
56,578
[1, 23810]
| |Car||she has a luxurious car.| |House||Biannca Raines has been known to live her own house.| |School||no verified
information found about school.| |College||No college found in our database.| |
7
* details on her educational history.. 9. Hassan tha jo
0.198079
0.445061
14
-55.867973
-55.911606
[22, 7414, 436, 1924, 25576, 7068, 396, 220, 37, 26, 106099, 31095, 8934, 200008]
mF_rl1
1
15,674
[0, 15674]
and 16.6% respectively. Newport Beach and Laguna Beach County Water District came squeaky close, saving 14.8% each. How? “
We are confident that some of the September savings for Newport Beach has to do with
0
* Our customers give us hope,” said Menifee, who has held water director
0.131601
0.362769
18
-54.113194
-53.717674
[22, 8721, 10872, 5290, 557, 11449, 4532, 3096, 11345, 2774, 81, 24, 2117, 947, 11624, 3427, 15983, 200008]
mF_rl1
1
15,674
[0, 15674]
and 16.6% respectively. Newport Beach and Laguna Beach County Water District came squeaky close, saving 14.8% each. How? “
We are confident that some of the September savings for Newport Beach has to do with
1
* A lot of hard conversations and stretching conversations with our customers over the last
0.087188
0.295276
16
-39.167206
-39.239708
[22, 352, 5693, 323, 5792, 49484, 341, 71954, 49484, 517, 1753, 10872, 1278, 290, 2954, 200008]
mF_rl1
1
15,674
[0, 15674]
and 16.6% respectively. Newport Beach and Laguna Beach County Water District came squeaky close, saving 14.8% each. How? “
We are confident that some of the September savings for Newport Beach has to do with
2
* It’s working,” Rindy said. “Public contact, awareness and education
0.143153
0.378355
17
-52.799068
-52.719276
[22, 1573, 1082, 6050, 4532, 423, 112394, 3096, 26, 1092, 18312, 6319, 24, 32962, 341, 10909, 200008]
mF_rl1
1
15,674
[0, 15674]
and 16.6% respectively. Newport Beach and Laguna Beach County Water District came squeaky close, saving 14.8% each. How? “
We are confident that some of the September savings for Newport Beach has to do with
3
* We asked our residents to reduce water
0.154174
0.39265
9
-14.707649
-14.768444
[22, 1665, 10151, 1753, 23504, 328, 9938, 3427, 200008]
End of preview. Expand in Data Studio

Muse Spark Open oracle-lens RL rollouts and training logs

Built with Muse Spark Open. Every rollout of the four released RL runs of the Muse Spark Open oracle-lens project (collection): the bullets the policies sampled at each of the 300 steps, the reconstruction reward and centred cosine each bullet earned, the sampler's and the trainer's log-probabilities, and the trainer's per-step metrics. Two native single-layer arms (fresh-init vs inverter-init SFT, identical prompts step for step), the multilayer lens, and the Qwen3.6-27B reader; 256 prompts x 8 samples per step = 614,400 bullets per run, nothing subsampled. Built 2026-10-26 from <run>/rollouts/step_XXXX.jsonl.gz, <run>/log.jsonl and <run>/rl_args.json on the muse-olens Modal volume; one row per sampled bullet; exact row counts in stats.json. The prompts are arbitrary high-entropy positions of web-text windows (not punctuation-only); readouts shown in the model cards come from the WorkspaceBench dataset's punctuation cells instead.

folder run what shipped checkpoints
native_lens/mF_rl1/ arm F: lr 6e-6 from the fresh-init SFT sft_F, reward = the clean reconstructor ar1c the arm-F release (held-out reconstruction 0.3307 at step 300) release_F, rl_F_step_0300 in ceselder/muse-spark-olens-lens
native_lens/mI_rl1/ arm I: lr 6e-6 from the inverter-init SFT sft_I (same prompts, step for step) the arm-I release (0.3296 at step 300) release_I, rl_I_step_0300
native_lens/mML_rl1/ multilayer lens: from mF_rl1 step 150, LoRA lr 6e-6 + per-layer affine input heads (lr 1e-5), one uniform layer in {8..56} per prompt, reward = ar1c + the cosine per-layer heads; its own pool (files 3..5, disjoint from the single-layer arms) mean per-layer held-out reconstruction 0.2734 at step 300 multilayer_step_0300
qwen_reader/xqm_rl_c/ Qwen3.6-27B reader, lr 6e-6 (map lr 2e-5) from sft_final, reward = ar1c, pool pool_wd8 the reader release (0.3269 at step 250) release, rl_step_0250 in ceselder/muse-spark-olens-qwen-reader

Columns

native_lens/<run>/rollouts.parquet (one row per sampled bullet; the source file had one row per prompt with 8 bullets):

column meaning
run, step run name, RL step (1-300)
row, src prompt id in the RL pool (src = JSON [pool file index, row] into /vol/xqm/pool_wd files 0..2 (mF_rl1, mI_rl1) or /vol/rl/pool_ml (mML_rl1): window-disjoint above-threshold positions of the rl-split documents, each used at one step only)
ctx_tail, true_next16 the last characters of the prompt's context and the document's actual next 16 tokens (decoded)
layer (multilayer run) the layer whose activation was read for that prompt (8 / 16 / 24 / 32 / 40 / 48 / 56)
g sample index within the group of 8
text the bullet (* ..., decoded from the sampled ids, capped at 16 content tokens)
reward, cos reconstruction reward clamp(cos, 0)^2 and centred cosine of the reconstructed bullet with the prompt's activation (the reconstructor service: ar1c; for the multilayer run the sampled layer's cosine head)
len generated tokens incl. * and the terminator (<= 18)
lp_engine, lp_train summed log-prob of the sample under the vLLM sampler (processed, under the grammar) and under the trainer's recomputation (the TIS pair; the rollout is one step off-policy by design)
ids JSON list of the sampled token ids (* 22, content, `<

qwen_reader/xqm_rl_c/rollouts.parquet: one row per sample as logged by xqm_rl.py (prompt row, the Qwen bullet, its Muse re-tokenised ids, reward, cos, log-probs). */train_log.parquet: the trainer's per-step scalars (log.jsonl: reward statistics, content NLL / entropy, KL to the SFT init, TIS log-ratio spread, held-out eval-greedy reconstruction every 25 steps, timings); rl_args.json: the exact arguments incl. the reward checkpoint and the reward-service self-test.

Licence

The bullets are outputs of Muse Spark Open (native arms) or of a Qwen3.6-27B reader of Muse activations; the prompt contexts are FineFineWeb text (m-a-p/FineFineWeb, its terms). Muse Model License (LICENSE.md + USE_POLICY.md, Meta's files verbatim) for the Muse-derived content; research use. Built with Muse Spark Open.

Provenance

src/muse_rl_train.py (native arms, --multilayer for mML_rl1) and src/xqm_rl.py (Qwen reader) of ceselder/muse-olens (commit 46b1c0b); conversion by scripts/hf_publish_muse.py::dataset_rollouts. Report: http://5.78.192.0/reports/view/muse-spark-olens/report.html.

Downloads last month
119

Collection including ceselder/muse-spark-olens-rl-rollouts