Jon-tow's workspace
Runs
1
Name
1 visualized
State
Notes
User
Tags
Created
Runtime
Sweep
batch_size
checkpoint_dir
checkpoint_interval
chunk_size
cliprange
cliprange_value
entity_name
epochs
eval_interval
gamma
gen_kwargs.do_sample
gen_kwargs.max_length
gen_kwargs.min_length
gen_kwargs.top_k
gen_kwargs.top_p
horizon
init_kl_coef
lam
learning_rate_init
learning_rate_target
lr_decay_steps
lr_ramp_steps
model_path
model_type
name
num_layers_unfrozen
num_rollouts
opt_betas
orchestrator
pipeline
ppo_epochs
project_name
seed
seq_length
target
tokenizer_path
total_steps
vf_coef
weight_decay
backward_time
exp_time
forward_time
generate_time
losses/pg_loss
Killed
-
jon-tow
25m 31s
-
128
ckpts
10000
128
0.2
0.2
jon-tow
1000
16
1
true
48
48
0
1
10000
0.2
0.95
0.0001412
0.0001412
79000
100
lvwerra/gpt2-imdb
AcceleratePPOModel
ppoconfig
2
128
[0.9,0.95]
PPOOrchestrator
PPOPipeline
4
ppo-test
1000
48
6
gpt2
10000
2.3
0.000001
0.16058
9.88318
0.11566
0.45808
-
1-1
of 1