Upload weborg_pol_rt_d24_s3500_red_all_adjacent_s1/pool_eval_topics.log with huggingface_hub
Browse files
weborg_pol_rt_d24_s3500_red_all_adjacent_s1/pool_eval_topics.log
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
W1005 07:16:02.684000 259342 torch/distributed/run.py:803]
|
| 2 |
+
W1005 07:16:02.684000 259342 torch/distributed/run.py:803] *****************************************
|
| 3 |
+
W1005 07:16:02.684000 259342 torch/distributed/run.py:803] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed.
|
| 4 |
+
W1005 07:16:02.684000 259342 torch/distributed/run.py:803] *****************************************
|
| 5 |
+
argv script=scripts.pool_eval args='-i base -g d24'
|
| 6 |
+
config source=base model_tag=d24 step=None device_batch_size=16 eval_tokens=8388608 threads=8 device_type='' run=dummy wandb_entity='' wandb_project=nanochat level=doc dataloader=bos_bestfit
|
| 7 |
+
Autodetected device type: cuda
|
| 8 |
+
2026-10-05 07:16:15,859 - nanochat.common - [32m[1mINFO[0m - Distributed world size: 8
|
| 9 |
+
2026-10-05 07:16:15,863 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Loading model from /mloscratch/homes/mljaggi/atlas-work/experiments/weborg_pol_rt_d24_s3500_red_all_adjacent_s1/d24/base with step 5568
|
| 10 |
+
2026-10-05 07:16:19,991 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Building model with config: {'sequence_len': 2048, 'vocab_size': 32768, 'n_layer': 24, 'n_head': 12, 'n_kv_head': 12, 'n_embd': 1536, 'window_pattern': 'SSSL'}
|
| 11 |
+
dataset weborg-topics-s3500: 24 pools, 32 steps x 262,144 tokens each
|
| 12 |
+
pool pool=adult level=doc bpb=1.205012 tokens=8388608
|
| 13 |
+
pool pool=art_design level=doc bpb=0.997128 tokens=8388608
|
| 14 |
+
pool pool=crime_law level=doc bpb=0.73815 tokens=8388608
|
| 15 |
+
pool pool=education_jobs level=doc bpb=0.825799 tokens=8388608
|
| 16 |
+
pool pool=entertainment level=doc bpb=0.994126 tokens=8388608
|
| 17 |
+
pool pool=fashion_beauty level=doc bpb=1.051614 tokens=8388608
|
| 18 |
+
pool pool=finance_business level=doc bpb=0.810547 tokens=8388608
|
| 19 |
+
pool pool=food_dining level=doc bpb=0.997557 tokens=8388608
|
| 20 |
+
pool pool=games level=doc bpb=1.07144 tokens=8388608
|
| 21 |
+
pool pool=hardware level=doc bpb=1.039778 tokens=8388608
|
| 22 |
+
pool pool=health level=doc bpb=0.852722 tokens=8388608
|
| 23 |
+
pool pool=history level=doc bpb=0.875064 tokens=8388608
|
| 24 |
+
pool pool=home_hobbies level=doc bpb=0.99004 tokens=8388608
|
| 25 |
+
pool pool=industrial level=doc bpb=0.903389 tokens=8388608
|
| 26 |
+
pool pool=literature level=doc bpb=0.96264 tokens=8388608
|
| 27 |
+
pool pool=politics level=doc bpb=0.799304 tokens=8388608
|
| 28 |
+
pool pool=religion level=doc bpb=0.895216 tokens=8388608
|
| 29 |
+
pool pool=science_tech level=doc bpb=0.894749 tokens=8388608
|
| 30 |
+
pool pool=social_life level=doc bpb=0.958852 tokens=8388608
|
| 31 |
+
pool pool=software level=doc bpb=0.933555 tokens=8388608
|
| 32 |
+
pool pool=software_dev level=doc bpb=1.05696 tokens=8388608
|
| 33 |
+
pool pool=sports_fitness level=doc bpb=0.981675 tokens=8388608
|
| 34 |
+
pool pool=transportation level=doc bpb=0.939597 tokens=8388608
|
| 35 |
+
pool pool=travel level=doc bpb=0.962945 tokens=8388608
|
| 36 |
+
summary model_tag=d24 source=base step=5568 dataset=weborg-topics-s3500 pools=24 dataloader=bos_bestfit level=doc mean_pool_bpb=0.947411 pool_bpb_spread=0.466862 mean_pool_bpb_chunk=nan eval_tokens_per_pool=8388608
|