Here's a list of grammarly scores:
Top-P | Score |
---|---|
0.59 | 68 |
0.60 | 82 |
0.61 | 97 |
0.62 | 99 |
0.63 | 97 |
0.64 | 81 |
0.65 | 97 |
0.66 | 99 |
0.67 | 96 |
0.68 | 99 |
0.69 | 99 |
0.70 | 99 |
0.71 | 96 |
0.72 | 99 |
0.73 | 99 |
0.74 | 99 |
0.75 | 86 |
0.76 | 85 |
0.77 | 93 |
0.78 | 99 |
0.79 | 97 |
0.80 | 96 |
0.81 | 93 |
0.82 | 92 |
0.83 | 91 |
0.84 | 96 |
0.85 | 76 |
0.86 | 87 |
0.87 | 70 |
0.88 | 86 |
0.89 | 87 |
0.90 | 79 |
Here is a list of length scores run through grammarly at 0.73 top_p: Somewhere along here I started playing with the context words. i.e. in the code: It adds "Response:" and "Reply:" before anything computed and before it was "Me:" + "You:".
Sentence Length | Score |
---|---|
Length 25 | 77 |
Length 24 | 80 |
Length 23 | 72 |
Length 22 | 71 |
Length 21 | 86 |
Length 20 | 88 |
Length 19 | 76 |
Length 18 | 68 |
Length 17 | 61 |
Length 16 | 53 |
Length 15 | 87 |
Length 14 | 83 |
Length 13 | 94 |
Length 12 | 55 |
Length 11 | 79 |
Length 10 | 99 |
Length 9 | 92 |
Length 8 | 85 |
Length 7 | 83 |
I'm unsure of how useful this would be. I think the input text was 5 words.