Views
No views yet
1perplexity: 167.37 seconds per pass - ETA 33.47 minutes - meta-405b-base - q8_0 - newest base was identical in bf16 and q8_0
2[1]1.3927,[2]1.6952,[3]1.5905,[4]1.4674,[5]1.3652,[6]1.3054,[7]1.2885,[8]1.2673,[9]1.2397,[10]1.2179,[11]1.2149,[12]1.2162,
3Final estimate: PPL = 1.2162 +/- 0.02128
4
5perplexity: 2197.87 seconds per pass - ETA 1 hours 49.88 minutes -- llama 405b - instruct - old BF16 -8head
6[1]2.1037,[2]2.4201,[3]2.0992,[4]1.8446,[5]1.6823,[6]1.5948,[7]1.5575,[8]1.5121,[9]1.4750,[10]1.4570,[11]1.4567,[12]1.4666,
7Final estimate: PPL = 1.4666 +/- 0.03184
8
9./llama-perplexity -m /scratch-10/lobotollama-q8_0.gguf -f wiki.test.raw -t 96 --chunks 12 -b 1024
10perplexity: 331.47 seconds per pass - ETA 33.13 minutes
11[1]2.6744,[2]3.4041,[3]2.9683,[4]2.8669,[5]2.7924,[6]2.7590,[7]2.8274,[8]2.8306,[9]2.7943,[10]2.7910,[11]2.8164,[12]2.9396,
12Final estimate: PPL = 2.9396 +/- 0.094971dtype: bfloat16
2merge_method: passthrough
3slices:
4- sources:
5 - layer_range: [0, 29]
6 model: /Meta-Llama-3.1-405B
7- sources:
8 - layer_range: [30, 35]
9 model: /Meta-Llama-3.1-405B
10- sources:
11 - layer_range: [36, 40]
12 model: /Meta-Llama-3.1-405B
13- sources:
14 - layer_range: [41, 45]
15 model: /Meta-Llama-3.1-405B
16- sources:
17 - layer_range: [46, 49]
18 model: /Meta-Llama-3.1-405B
19- sources:
20 - layer_range: [50, 54]
21 model: /Meta-Llama-3.1-405B
22- sources:
23 - layer_range: [55, 59]
24 model: /Meta-Llama-3.1-405B
25- sources:
26 - layer_range: [60, 64]
27 model: /Meta-Llama-3.1-405B
28- sources:
29 - layer_range: [65, 69]
30 model: /Meta-Llama-3.1-405B
31- sources:
32 - layer_range: [70, 74]
33 model: /Meta-Llama-3.1-405B
34- sources:
35 - layer_range: [75, 79]
36 model: /Meta-Llama-3.1-405B
37- sources:
38 - layer_range: [80, 84]
39 model: /Meta-Llama-3.1-405B
40- sources:
41 - layer_range: [85, 126]
42 model: /Meta-Llama-3.1-405B