Hugging Face's logo

tangledgroup
/
tangled-alpha-0.2-core

Text Generation
Transformers
chat
core
base
instruct
reason
conversational
Model card Files Community
tangled-alpha-0.2-core / README.md
mtasic85's picture
mtasic85
readme
cdeea60
preview code
|
raw
history blame contribute delete
5.31 kB
metadata
license: mit
pipeline_tag: text-generation
library_name: transformers
language:
  - en
  - am
  - ar
  - as
  - az
  - be
  - bg
  - bn
  - br
  - bs
  - ca
  - cs
  - cy
  - da
  - de
  - el
  - eo
  - es
  - et
  - eu
  - fa
  - ff
  - fi
  - fr
  - fy
  - ga
  - gd
  - gl
  - gn
  - gu
  - ha
  - he
  - hi
  - hr
  - ht
  - hu
  - hy
  - id
  - ig
  - is
  - it
  - ja
  - jv
  - ka
  - kk
  - km
  - kn
  - ko
  - ku
  - ky
  - la
  - lg
  - li
  - ln
  - lo
  - lt
  - lv
  - mg
  - mk
  - ml
  - mn
  - mr
  - ms
  - my
  - ne
  - nl
  - 'no'
  - ns
  - om
  - or
  - pa
  - pl
  - ps
  - pt
  - qu
  - rm
  - ro
  - ru
  - sa
  - si
  - sc
  - sd
  - sk
  - sl
  - so
  - sq
  - sr
  - ss
  - su
  - sv
  - sw
  - ta
  - te
  - th
  - tl
  - tn
  - tr
  - ug
  - uk
  - ur
  - uz
  - vi
  - wo
  - xh
  - yi
  - yo
  - zu
datasets:
  - ontocord/fineweb-permissive-multilingual-2m
  - distily/c4_multilingual_1M
  - data-silence/sumnews
  - xu-song/cc100-samples
  - badrex/llm-emoji-dataset
  - fblgit/simple-math
  - Gusarich/math-expressions-1m
  - neuralwork/arxiver
  - christopher/rosetta-code
  - nampdn-ai/tiny-codes
  - JeanKaddour/minipile
  - NousResearch/hermes-function-calling-v1
  - simplescaling/s1K-1.1
  - mlabonne/open-perfectblend
  - allenai/tulu-3-sft-mixture
  - rombodawg/Everything_Instruct_Multilingual
  - open-r1/OpenR1-Math-220k
  - open-thoughts/OpenThoughts-114k
  - cognitivecomputations/dolphin-r1
  - simplescaling/s1K-1.1
tags:
  - chat
  - core
  - base
  - instruct
  - reason

tangled-alpha-0.2-core

logo

time python -B prepare_core_datasets.py

Progress: 100%|████████| 220/220 [23:15<00:00,  6.34s/it]
Workers are finished.██| 220/220 [23:15<00:00,  6.34s/it]
Finished data processing!
i=0, block_size=8192, chunk_size=16384000, len(dataset)=893355, len(dataset) * block_size=7318364160
Total number of tokens in the optimized dataset '../core-data-0-8192-2000' is 7318364160

CUDA_VISIBLE_DEVICES=0 CUDA_LAUNCH_BLOCKING=0 PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True litgpt pretrain --config pretrain-core-model.yaml

Seed set to 23
Time to instantiate model: 0.32 seconds.
Total parameters: 217,088,512
Verifying settings ...
Measured TFLOPs: 3548.40

Epoch 1 | iter 256 step 1 | loss train: 11.716, val: n/a | iter time: 1735.26 ms (step) remaining time: 4 days, 11:06:29
Epoch 1 | iter 512 step 2 | loss train: 11.534, val: n/a | iter time: 1102.77 ms (step) remaining time: 4 days, 2:31:30
Epoch 1 | iter 768 step 3 | loss train: 11.356, val: n/a | iter time: 1095.87 ms (step) remaining time: 3 days, 23:44:12
Epoch 1 | iter 1024 step 4 | loss train: 11.162, val: n/a | iter time: 1099.92 ms (step) remaining time: 3 days, 22:18:27
Epoch 1 | iter 1280 step 5 | loss train: 11.018, val: n/a | iter time: 1096.45 ms (step) remaining time: 3 days, 21:24:35
Epoch 1 | iter 1536 step 6 | loss train: 10.901, val: n/a | iter time: 1093.65 ms (step) remaining time: 3 days, 20:48:11
Epoch 1 | iter 1792 step 7 | loss train: 10.850, val: n/a | iter time: 1100.16 ms (step) remaining time: 3 days, 20:22:00
Epoch 1 | iter 2048 step 8 | loss train: 10.780, val: n/a | iter time: 1092.67 ms (step) remaining time: 3 days, 20:01:57
Epoch 1 | iter 2304 step 9 | loss train: 10.692, val: n/a | iter time: 1095.77 ms (step) remaining time: 3 days, 19:45:57
Epoch 1 | iter 2560 step 10 | loss train: 10.678, val: n/a | iter time: 1092.12 ms (step) remaining time: 3 days, 19:32:43
Epoch 1 | iter 2816 step 11 | loss train: 10.619, val: n/a | iter time: 1094.44 ms (step) remaining time: 3 days, 19:21:32
Epoch 1 | iter 3072 step 12 | loss train: 10.588, val: n/a | iter time: 1102.51 ms (step) remaining time: 3 days, 19:12:30
Epoch 1 | iter 3328 step 13 | loss train: 10.514, val: n/a | iter time: 1095.57 ms (step) remaining time: 3 days, 19:04:07
Epoch 1 | iter 3584 step 14 | loss train: 10.472, val: n/a | iter time: 1104.00 ms (step) remaining time: 3 days, 18:56:56
Epoch 1 | iter 3840 step 15 | loss train: 10.431, val: n/a | iter time: 1096.00 ms (step) remaining time: 3 days, 18:50:21
Epoch 1 | iter 4096 step 16 | loss train: 10.392, val: n/a | iter time: 1098.34 ms (step) remaining time: 3 days, 18:44:25
Epoch 1 | iter 4352 step 17 | loss train: 10.360, val: n/a | iter time: 1106.53 ms (step) remaining time: 3 days, 18:38:58
Epoch 1 | iter 4608 step 18 | loss train: 10.329, val: n/a | iter time: 1084.95 ms (step) remaining time: 3 days, 18:33:58
Epoch 1 | iter 4864 step 19 | loss train: 10.296, val: n/a | iter time: 1096.22 ms (step) remaining time: 3 days, 18:29:12
Epoch 1 | iter 5120 step 20 | loss train: 10.236, val: n/a | iter time: 1093.39 ms (step) remaining time: 3 days, 18:24:51
# ...

Backup wandb:

mv wandb wandb-pretrain-core

Chat with model:

CUDA_VISIBLE_DEVICES=0 CUDA_LAUNCH_BLOCKING=0 PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True litgpt chat ../out/pretrain-core/final

CUDA_VISIBLE_DEVICES=0 CUDA_LAUNCH_BLOCKING=0 PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True time litgpt evaluate --tasks 'leaderboard' --out_dir '../evaluate/pretrain-core/leaderboard/' --batch_size 1 --dtype 'bfloat16' '../out/pretrain-core/final'

# ...