pszemraj commited on
Commit
ded3a22
1 Parent(s): ae142c5

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +83 -9
README.md CHANGED
@@ -2,17 +2,84 @@
2
  license: apache-2.0
3
  tags:
4
  - generated_from_trainer
 
 
5
  metrics:
6
  - rouge
7
  model-index:
8
- - name: flan-t5-base-fleece2instructions-r1
9
- results: []
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10
  ---
11
 
12
- <!-- This model card has been generated automatically according to the information the Trainer had access to. You
13
- should probably proofread and complete it, then remove this comment. -->
14
 
15
- # flan-t5-base-fleece2instructions-r1
 
 
16
 
17
  This model is a fine-tuned version of [google/flan-t5-base](https://huggingface.co/google/flan-t5-base) on the None dataset.
18
  It achieves the following results on the evaluation set:
@@ -23,13 +90,20 @@ It achieves the following results on the evaluation set:
23
  - Rougelsum: 56.9171
24
  - Gen Len: 13.1493
25
 
26
- ## Model description
27
 
28
- More information needed
29
 
30
- ## Intended uses & limitations
 
 
 
 
 
 
 
 
31
 
32
- More information needed
33
 
34
  ## Training and evaluation data
35
 
 
2
  license: apache-2.0
3
  tags:
4
  - generated_from_trainer
5
+ datasets:
6
+ - pszemraj/fleece2instructions
7
  metrics:
8
  - rouge
9
  model-index:
10
+ - name: flan-t5-base-instructiongen
11
+ results:
12
+ - task:
13
+ name: Sequence-to-sequence Language Modeling
14
+ type: text2text-generation
15
+ dataset:
16
+ name: pszemraj/fleece2instructions
17
+ type: pszemraj/fleece2instructions
18
+ split: validation
19
+ metrics:
20
+ - name: Rouge1
21
+ type: rouge
22
+ value: 58.9516
23
+ widget:
24
+ - text: >-
25
+ You'll need to start by choosing the right venue. Consider the type of
26
+ atmosphere and the size of the area that will be suitable for the number
27
+ of guests you plan to invite. Choose the right decorations based on your
28
+ brother's interests, such as balloons in his favorite colors, banners, and
29
+ streamers. Next, decide on the food and drinks, making sure they are tasty
30
+ and appropriate for the occasion. Then decide on the other games, music,
31
+ and entertainment that will make the party memorable. Finally, involve
32
+ your brother's friends and family to help create the perfect surprise.
33
+ example_title: birthday party
34
+ - text: 1) cookies and cream 2) chocolate chip 3) mint chip 4) oreo
35
+ example_title: ice cream
36
+ - text: >-
37
+ Start by selecting a scale model of a building that fits the theme. Use a
38
+ hobby knife and glue to cut and assemble the model into a ruined or
39
+ abandoned version of itself, adding details like broken windows and
40
+ graffiti. Create a base for the diorama using foam, plaster, or other
41
+ materials, and paint it to resemble a ruined street or sidewalk. Add
42
+ miniature vehicles, debris, and figures to complete the scene, and use
43
+ weathering techniques like dry brushing and rust washes to add realism.
44
+ Display the diorama in a shadow box or other protective case to showcase
45
+ your work.
46
+ example_title: Miniature diorama creation
47
+ - text: >-
48
+ Start by selecting clothing that is futuristic and edgy, such as leather
49
+ jackets, neon-colored accessories, and tech-inspired patterns. Add
50
+ accessories like goggles, cybernetic implants, and LED lights to enhance
51
+ the cyberpunk vibe. Use makeup and body paint to create a futuristic look,
52
+ such as metallic skin or neon makeup. Consider adding functional elements
53
+ to your costume, such as a built-in backpack or hidden pockets for your
54
+ tech gadgets. Finally, practice your confident walk and embrace your inner
55
+ cyberpunk for a memorable and immersive costume experience.
56
+ example_title: Cyberpunk costume design
57
+ - text: >-
58
+ Start by creating a base terrain with mountains, valleys, and other
59
+ natural features. Use fractal noise and displacement mapping to add
60
+ texture and detail to the terrain, and experiment with different materials
61
+ like rock, grass, and water. Add surreal elements like floating islands,
62
+ giant mushrooms, or impossible geometry to create a dreamlike atmosphere.
63
+ Use lighting and color grading to enhance the mood and tone of the scene,
64
+ and render the final image at a high resolution for maximum impact. Share
65
+ your surreal landscape with the world and inspire others to explore the
66
+ possibilities of 3D art.
67
+ example_title: Surreal 3D landscape creation
68
+ - text: >-
69
+ Start by setting a realistic goal and creating a training plan. Build up
70
+ your mileage gradually over time, and incorporate cross-training and
71
+ strength exercises to prevent injury and improve endurance. Be sure to
72
+ stay hydrated and properly fuel your body with nutritious foods. Listen to
73
+ your body and adjust your training as needed to avoid overexertion or
74
+ burnout. Finally, taper your training in the weeks leading up to the race
75
+ to give your body time to rest and recover before the big day.
76
+ example_title: Marathon training
77
  ---
78
 
 
 
79
 
80
+ # flan-t5-base-instructiongen
81
+
82
+ Instead of generating questions from text, generate instructions for LLMs!
83
 
84
  This model is a fine-tuned version of [google/flan-t5-base](https://huggingface.co/google/flan-t5-base) on the None dataset.
85
  It achieves the following results on the evaluation set:
 
90
  - Rougelsum: 56.9171
91
  - Gen Len: 13.1493
92
 
93
+ ## Intended uses & limitations
94
 
95
+ > Of the three models fine-tuned so far, `flan-t5-base` is in an awkward position where it has the largest model file size, but not the best performance. I'd recommend looking at the two linked below.
96
 
97
+ This is just a `base` FLAN model, and is mostly uploaded for comparison with the [FLAN-small](https://huggingface.co/pszemraj/flan-t5-small-instructiongen) and [bart-base](https://huggingface.co/pszemraj/bart-base-instructiongen) variants.
98
+
99
+ Additionally, it was trained on a dataset of **only** instructions+outputs, with the `inputs` filtered out. This means that text of *1) cookies and cream 2) chocolate chip 3) mint chip 4) oreo* will **not** get you *"Rank the following ice cream flavors: oreo, mint chip, chocolate chip, cookies and cream"*
100
+
101
+ ## Training and evaluation data
102
+
103
+ See the linked dataset `pszemraj/fleece2instructions` - it is a filtered/formatted version of `tatsu-lab/alpaca` to generate instructions for arbitrary text.
104
+
105
+ - Some of the API examples are intentionally weird to demonstrate the generalizability of the model.
106
 
 
107
 
108
  ## Training and evaluation data
109