Bilalelbouardi's workspace
Runs
7
State
Notes
User
Tags
Created
Runtime
Sweep
actor_rollout_ref.actor.checkpoint.contents
actor_rollout_ref.actor.clip_ratio
actor_rollout_ref.actor.clip_ratio_c
actor_rollout_ref.actor.clip_ratio_high
actor_rollout_ref.actor.clip_ratio_low
actor_rollout_ref.actor.entropy_coeff
actor_rollout_ref.actor.fsdp_config.fsdp_size
actor_rollout_ref.actor.fsdp_config.offload_policy
actor_rollout_ref.actor.fsdp_config.optimizer_offload
actor_rollout_ref.actor.fsdp_config.param_offload
actor_rollout_ref.actor.fsdp_config.reshard_after_forward
actor_rollout_ref.actor.fsdp_config.wrap_policy.min_num_params
actor_rollout_ref.actor.grad_clip
actor_rollout_ref.actor.kl_loss_coef
actor_rollout_ref.actor.kl_loss_type
actor_rollout_ref.actor.loss_agg_mode
actor_rollout_ref.actor.optim.lr
actor_rollout_ref.actor.optim.lr_warmup_steps
actor_rollout_ref.actor.optim.lr_warmup_steps_ratio
actor_rollout_ref.actor.optim.total_training_steps
actor_rollout_ref.actor.optim.warmup_style
actor_rollout_ref.actor.optim.weight_decay
actor_rollout_ref.actor.ppo_epochs
actor_rollout_ref.actor.ppo_max_token_len_per_gpu
actor_rollout_ref.actor.ppo_micro_batch_size_per_gpu
actor_rollout_ref.actor.ppo_mini_batch_size
actor_rollout_ref.actor.shuffle
actor_rollout_ref.actor.strategy
actor_rollout_ref.actor.ulysses_sequence_parallel_size
actor_rollout_ref.actor.use_dynamic_bsz
actor_rollout_ref.actor.use_kl_loss
actor_rollout_ref.actor.use_torch_compile
actor_rollout_ref.hybrid_engine
actor_rollout_ref.model.enable_gradient_checkpointing
actor_rollout_ref.model.path
actor_rollout_ref.model.trust_remote_code
actor_rollout_ref.model.use_liger
actor_rollout_ref.model.use_remove_padding
actor_rollout_ref.ref.fsdp_config.param_offload
actor_rollout_ref.ref.fsdp_config.reshard_after_forward
actor_rollout_ref.ref.fsdp_config.wrap_policy.min_num_params
actor_rollout_ref.ref.log_prob_max_token_len_per_gpu
actor_rollout_ref.ref.log_prob_micro_batch_size_per_gpu
actor_rollout_ref.ref.log_prob_use_dynamic_bsz
Crashed
reason-wang
12h 47m 1s
-
["model","optimizer","extra"]
0.2
3
0.2
0.2
0.001
-1
false
true
true
true
0
1
0.001
mse
token-mean
5.0000e-7
-1
0
200
constant
0.01
1
16384
2
512
false
fsdp
1
false
true
true
true
false
Qwen/Qwen2.5-VL-3B-Instruct
true
false
false
true
true
0
16384
4
false
Crashed
reason-wang
23h 25m 55s
-
["model","optimizer","extra"]
0.2
3
0.2
0.2
0.01
-1
false
true
true
true
0
1
0.001
mse
token-mean
4.0000e-7
-1
0
200
constant
0.01
1
16384
1
4
false
fsdp
1
false
true
true
true
false
Qwen/Qwen2.5-VL-3B-Instruct
false
false
false
true
true
0
16384
4
false
Crashed
reason-wang
2h 58m 5s
-
["model","optimizer","extra"]
0.2
3
0.2
0.2
0.001
-1
false
true
true
true
0
1
0.001
mse
token-mean
5.0000e-7
-1
0
200
constant
0.01
1
16384
2
512
false
fsdp
1
false
true
true
true
false
Qwen/Qwen2.5-VL-3B-Instruct
true
false
false
true
true
0
16384
4
false
Crashed
reason-wang
11h 2m 44s
-
["model","optimizer","extra"]
0.2
3
0.2
0.2
0.005
-1
false
true
true
true
0
1
0.005
mse
token-mean
4.0000e-7
-1
0
200
constant
0.01
1
16384
2
32
false
fsdp
1
false
true
true
true
false
Qwen/Qwen2.5-7B-Instruct
false
false
false
true
true
0
16384
4
false
Crashed
reason-wang
11h 30m 16s
-
["model","optimizer","extra"]
0.2
3
0.2
0.2
0.0001
-1
false
true
true
true
0
1
0.001
mse
token-mean
4.0000e-7
-1
0
200
constant
0.01
1
16384
2
128
false
fsdp
1
false
true
true
true
false
Qwen/Qwen2.5-3B-Instruct
false
false
false
true
true
0
16384
4
false
Crashed
reason-wang
10h 13m 16s
-
["model","optimizer","extra"]
0.2
3
0.2
0.2
0.001
-1
false
true
true
true
0
1
0.001
mse
token-mean
5.0000e-7
-1
0
100
constant
0.01
1
16384
2
128
false
fsdp
1
false
true
true
true
false
Qwen/Qwen2.5-3B-Instruct
false
false
false
true
true
0
16384
4
false
Crashed
reason-wang
7h 41m 1s
-
["model","optimizer","extra"]
0.2
3
0.2
0.2
0.001
-1
false
true
true
true
0
1
0.001
mse
token-mean
5.0000e-7
-1
0
200
constant
0.01
1
16384
2
64
false
fsdp
1
false
true
true
true
false
Qwen/Qwen2.5-3B-Instruct
false
false
false
true
true
0
16384
4
false
1-7
of 7