Yuya-yamamoto's workspace
Runs
1
Name
1 visualized
State
Notes
User
Tags
Created
Runtime
Sweep
_name_or_path
adafactor
adam_beta1
adam_beta2
adam_epsilon
add_cross_attention
architectures
attention_dropout
auto_find_batch_size
bf16
bf16_full_eval
bos_token_id
chunk_size_feed_forward
data_seed
dataloader_drop_last
dataloader_num_workers
dataloader_pin_memory
ddp_bucket_cap_mb
ddp_find_unused_parameters
ddp_timeout
debug
deepspeed
disable_tqdm
diversity_penalty
do_eval
do_predict
do_sample
do_train
early_stopping
encoder_no_repeat_ngram_size
eos_token_id
eval_accumulation_steps
eval_batch_size
eval_delay
eval_steps
evaluation_strategy
fp16
fp16_backend
fp16_full_eval
fp16_opt_level
fsdp
fsdp_config
fsdp_min_num_params
fsdp_transformer_layer_cls_to_wrap
Finished
-
yuya-yamamoto
4d 5h 13m 3s
-
abeja/gpt-neox-japanese-2.7b
false
0.9
0.999
1.0000e-8
false
["GPTNeoXJapaneseForCausalLM"]
0.1
false
false
false
31999
0
None
false
0
true
None
None
1800
[]
None
false
0
true
false
false
false
false
0
31999
None
4
0
1000
steps
true
auto
false
O1
[]
{'fsdp_min_num_params': 0, 'xla': False, 'xla_fsdp_grad_ckpt': False}
0
None
1-1
of 1