Jqi's workspace
Runs
6
Name
2 visualized
State
Notes
User
Tags
Created
Runtime
Sweep
cfg/activation
cfg/answer_only_loss
cfg/apex_transformer_log_level
cfg/apply_query_key_layer_scaling
cfg/attention_dropout
cfg/attention_type
cfg/bias
cfg/bias_activation_fusion
cfg/bias_dropout_add_fusion
cfg/data/chat
cfg/data/train_ds/add_bos
cfg/data/train_ds/add_eos
cfg/data/train_ds/add_sep
cfg/data/train_ds/concat_sampling_probabilities
cfg/data/train_ds/context_key
cfg/data/train_ds/drop_last
cfg/data/train_ds/file_names
cfg/data/train_ds/global_batch_size
cfg/data/train_ds/label_key
cfg/data/train_ds/max_seq_length
cfg/data/train_ds/micro_batch_size
cfg/data/train_ds/min_seq_length
cfg/data/train_ds/num_workers
cfg/data/train_ds/pin_memory
cfg/data/train_ds/separate_prompt_and_response_with_newline
cfg/data/train_ds/shuffle
cfg/data/train_ds/truncation_field
cfg/data/validation_ds/add_bos
cfg/data/validation_ds/add_eos
cfg/data/validation_ds/add_sep
cfg/data/validation_ds/context_key
cfg/data/validation_ds/drop_last
cfg/data/validation_ds/file_names
cfg/data/validation_ds/global_batch_size
cfg/data/validation_ds/label_key
cfg/data/validation_ds/max_seq_length
cfg/data/validation_ds/metric/name
cfg/data/validation_ds/micro_batch_size
cfg/data/validation_ds/min_seq_length
cfg/data/validation_ds/num_workers
cfg/data/validation_ds/pin_memory
cfg/data/validation_ds/separate_prompt_and_response_with_newline
cfg/data/validation_ds/shuffle
cfg/data/validation_ds/truncation_field
Finished
-
jqi
1h 52s
-
fast-swiglu
true
30
true
0
multihead
false
false
false
false
false
true
false
[1]
input
true
["/home/sa/jqi/alchemy/nemo/sft/data/dolly-15k.jsonl"]
128
output
1024
4
1
4
true
false
false
context
false
true
false
input
true
["/home/sa/jqi/alchemy/nemo/sft/data/dolly-15k.jsonl"]
128
output
1024
loss
4
1
4
true
false
false
context
Finished
-
jqi
1h 45s
-
fast-swiglu
true
30
true
0
multihead
false
false
false
false
false
true
false
[1]
input
true
["/home/sa/jqi/alchemy/nemo/sft/data/dolly-15k.jsonl"]
128
output
1024
4
1
4
true
false
false
context
false
true
false
input
true
["/home/sa/jqi/alchemy/nemo/sft/data/dolly-15k.jsonl"]
128
output
1024
loss
4
1
4
true
false
false
context
Finished
-
jqi
23m 53s
-
fast-swiglu
true
30
true
0
multihead
false
false
false
false
false
true
false
[1]
input
true
["/lustre/fsw/sa/daochens/hkust/share/hkust/nemo-fw/datasets/dolly-15k-nemo/dolly-15k.jsonl"]
128
output
1024
4
1
4
true
false
true
context
false
true
false
input
true
["/lustre/fsw/sa/daochens/hkust/share/hkust/nemo-fw/datasets/dolly-15k-nemo/dolly-15k.jsonl"]
128
output
1024
loss
4
1
4
true
false
false
context
Crashed
-
jqi
24m 1s
-
fast-swiglu
true
30
true
0
multihead
false
false
false
false
false
true
false
[1]
input
true
["/lustre/fsw/sa/daochens/hkust/share/hkust/nemo-fw/datasets/dolly-15k-nemo/dolly-15k.jsonl"]
128
output
900
4
1
4
true
false
true
context
false
true
false
input
true
["/lustre/fsw/sa/daochens/hkust/share/hkust/nemo-fw/datasets/dolly-15k-nemo/dolly-15k.jsonl"]
128
output
900
loss
4
1
4
true
false
false
context
Crashed
-
jqi
23m 31s
-
fast-swiglu
true
30
true
0
multihead
false
false
false
false
false
true
false
[1]
input
true
["/lustre/fsw/sa/daochens/hkust/share/hkust/nemo-fw/datasets/dolly-15k-nemo/dolly-15k.jsonl"]
128
output
1024
4
1
4
true
false
true
context
false
true
false
input
true
["/lustre/fsw/sa/daochens/hkust/share/hkust/nemo-fw/datasets/dolly-15k-nemo/dolly-15k.jsonl"]
128
output
1024
loss
4
1
4
true
false
false
context
Finished
-
jqi
20m 32s
-
fast-swiglu
true
30
true
0
multihead
false
false
false
false
false
true
false
[1]
input
true
["/lustre/fsw/sa/daochens/hkust/share/hkust/nemo-fw/datasets/dolly-15k-nemo/dolly-15k.jsonl"]
128
output
1024
4
1
4
true
false
true
context
false
true
false
input
true
["/lustre/fsw/sa/daochens/hkust/share/hkust/nemo-fw/datasets/dolly-15k-nemo/dolly-15k.jsonl"]
128
output
1024
loss
4
1
4
true
false
false
context
1-6
of 6