Update README.md
Browse files
README.md
CHANGED
@@ -31,6 +31,55 @@ Just like my teacher gave me homework, i thought maybe we can also add some of t
|
|
31 |
|
32 |
It was created with very simple code that is in the repo, if you add more complex operations and so.. please share the code :D thank you
|
33 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
34 |
Note to contributors:
|
35 |
**thank you to those contributing on the experiment with beautiful commits and good spirit**
|
36 |
* The model needs some splits
|
@@ -39,10 +88,23 @@ Note to contributors:
|
|
39 |
* Lets aim to build an ablation & paper together. All contributors will be cited.
|
40 |
* Add your log entry on the version so we can keep a track, thanks.
|
41 |
|
42 |
-
|
43 |
## Versions
|
|
|
44 |
24.01.24 Added gradual complexity on a separate script
|
45 |
20-23.01.24 Multiple contributions with operations and increased complexity on the main generator script.
|
|
|
46 |
|
|
|
47 |
If you use Simple Math o train your model, please cite on the modelcard or the paper.
|
48 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
31 |
|
32 |
It was created with very simple code that is in the repo, if you add more complex operations and so.. please share the code :D thank you
|
33 |
|
34 |
+
## Does it Works?
|
35 |
+
### 34BEAGLES Evaluation:
|
36 |
+
```
|
37 |
+
hf (pretrained=/data/models/UNA-34Beagles-v1-final,dtype=bfloat16,trust_remote_code=True), gen_kwargs: (None), limit: None, num_fewshot: None, batch_size: auto (8)
|
38 |
+
| Tasks |Version|Filter|n-shot| Metric |Value | |Stderr|
|
39 |
+
|--------------|-------|------|-----:|--------|-----:|---|-----:|
|
40 |
+
|arc_challenge |Yaml |none | 25|acc |0.7039|± |0.0133|
|
41 |
+
| | |none | 25|acc_norm|0.7321|± |0.0129|
|
42 |
+
|truthfulqa_mc2|Yaml |none | 0|acc |0.7387|± |0.0141|
|
43 |
+
|
44 |
+
hf (pretrained=/data/models/UNA-34Beagles-v1-final,dtype=bfloat16,trust_remote_code=True), gen_kwargs: (None), limit: None, num_fewshot: None, batch_size: auto
|
45 |
+
|Tasks|Version| Filter |n-shot| Metric |Value | |Stderr|
|
46 |
+
|-----|-------|----------|-----:|-----------|-----:|---|-----:|
|
47 |
+
|gsm8k|Yaml |get-answer| 5|exact_match|0.6399|± |0.0132|
|
48 |
+
|
49 |
+
| Groups |Version|Filter|n-shot|Metric|Value | |Stderr|
|
50 |
+
|------------------|-------|------|-----:|------|-----:|---|-----:|
|
51 |
+
|mmlu |N/A |none | 0|acc |0.7477|± |0.1079|
|
52 |
+
| - humanities |N/A |none | 0|acc |0.7188|± |0.0855|
|
53 |
+
| - other |N/A |none | 0|acc |0.7950|± |0.1057|
|
54 |
+
| - social_sciences|N/A |none | 0|acc |0.8297|± |0.0664|
|
55 |
+
| - stem |N/A |none | 0|acc |0.6641|± |0.1291|
|
56 |
+
```
|
57 |
+
|
58 |
+
### 34BEAGLES-MATH Evaluation
|
59 |
+
```
|
60 |
+
hf (pretrained=/data/models/34BeaglesMath-v1,dtype=bfloat16,trust_remote_code=True), gen_kwargs: (None), limit: None, num_fewshot: None, batch_size: auto
|
61 |
+
|Tasks|Version| Filter |n-shot| Metric |Value | |Stderr|
|
62 |
+
|-----|-------|----------|-----:|-----------|-----:|---|-----:|
|
63 |
+
|gsm8k|Yaml |get-answer| 5|exact_match|0.6505|± |0.0131|
|
64 |
+
|
65 |
+
hf (pretrained=/data/models/34BeaglesMath-v1,dtype=bfloat16,trust_remote_code=True), gen_kwargs: (None), limit: None, num_fewshot: None, batch_size: auto (8)
|
66 |
+
| Tasks |Version|Filter|n-shot| Metric |Value | |Stderr|
|
67 |
+
|--------------|-------|------|-----:|--------|-----:|---|-----:|
|
68 |
+
|arc_challenge |Yaml |none | 25|acc |0.7090|± |0.0133|
|
69 |
+
| | |none | 25|acc_norm|0.7329|± |0.0129|
|
70 |
+
|truthfulqa_mc2|Yaml |none | 0|acc |0.7378|± |0.0141|
|
71 |
+
|
72 |
+
| Groups |Version|Filter|n-shot|Metric|Value | |Stderr|
|
73 |
+
|------------------|-------|------|-----:|------|-----:|---|-----:|
|
74 |
+
|mmlu |N/A |none | 0|acc |0.7524|± |0.1045|
|
75 |
+
| - humanities |N/A |none | 0|acc |0.7307|± |0.0846|
|
76 |
+
| - other |N/A |none | 0|acc |0.7937|± |0.1029|
|
77 |
+
| - social_sciences|N/A |none | 0|acc |0.8274|± |0.0667|
|
78 |
+
| - stem |N/A |none | 0|acc |0.6708|± |0.1236|
|
79 |
+
```
|
80 |
+
|
81 |
+
I think it works. Not too difficult, not too easy, as a curriculum works well :)
|
82 |
+
|
83 |
Note to contributors:
|
84 |
**thank you to those contributing on the experiment with beautiful commits and good spirit**
|
85 |
* The model needs some splits
|
|
|
88 |
* Lets aim to build an ablation & paper together. All contributors will be cited.
|
89 |
* Add your log entry on the version so we can keep a track, thanks.
|
90 |
|
91 |
+
|
92 |
## Versions
|
93 |
+
```
|
94 |
24.01.24 Added gradual complexity on a separate script
|
95 |
20-23.01.24 Multiple contributions with operations and increased complexity on the main generator script.
|
96 |
+
```
|
97 |
|
98 |
+
## Citations
|
99 |
If you use Simple Math o train your model, please cite on the modelcard or the paper.
|
100 |
+
|
101 |
+
```
|
102 |
+
@misc{simplemath,
|
103 |
+
title={Simple-Math: 2+2=4 4-1=3},
|
104 |
+
author={Xavier Murias},
|
105 |
+
year={2024},
|
106 |
+
publisher = {Juanako.AI},
|
107 |
+
journal = {HuggingFace repository},
|
108 |
+
howpublished = {\url{https://huggingface.co/datasets/fblgit/simple-math}},
|
109 |
+
}
|
110 |
+
```
|