Llychinalz's workspace
Runs
7
Name
3 visualized
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_checkpointing
actor_rollout_ref.actor.entropy_coeff
actor_rollout_ref.actor.entropy_from_logits_with_chunking
actor_rollout_ref.actor.fsdp_config.forward_prefetch
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.min_lr_ratio
actor_rollout_ref.actor.optim.num_cycles
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_mini_batch_size
actor_rollout_ref.actor.quant_cross_entropy_coeff
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_activation_offload
actor_rollout_ref.model.enable_gradient_checkpointing
actor_rollout_ref.model.fused_kernel_options.impl_backend
actor_rollout_ref.model.lora_alpha
actor_rollout_ref.model.lora_rank
actor_rollout_ref.model.path
Finished
-
llychinalz
2s
-
["model","optimizer","extra"]
0.2
10
0.28
0.2
false
0
false
false
-1
false
true
true
true
0
1
0
low_var_kl
token-mean
0.000001
10
0
0
0.5
1100
constant
0.1
1
22528
32
0
false
fsdp
8
true
false
true
true
false
true
torch
16
0
/workspace/Qwen2.5-32B
Finished
-
llychinalz
1s
-
["model","optimizer","extra"]
0.2
10
0.28
0.2
false
0
false
false
-1
false
true
true
true
0
1
0
low_var_kl
token-mean
0.000001
10
0
0
0.5
1100
constant
0.1
1
45056
32
0
false
fsdp
1
true
false
true
true
false
true
torch
16
0
/workspace/DeepSeek-R1-Distill-Qwen-1.5B
Finished
-
llychinalz
1s
-
["model","optimizer","extra"]
0.2
10
0.28
0.2
false
0
false
false
-1
false
true
true
true
0
1
0
low_var_kl
token-mean
0.000001
10
0
0
0.5
1100
constant
0.1
1
45056
32
0
false
fsdp
1
true
false
true
true
false
true
torch
16
0
/workspace/DeepSeek-R1-Distill-Qwen-1.5B
Finished
-
llychinalz
2s
-
["model","optimizer","extra"]
0.2
10
0.28
0.2
false
0
false
false
-1
false
true
true
true
0
1
0
low_var_kl
token-mean
0.000001
10
0
0
0.5
1100
constant
0.1
1
22528
32
0
false
fsdp
8
true
false
true
true
false
true
torch
16
0
/root/Qwen2.5-32B
Finished
-
llychinalz
1s
-
["model","optimizer","extra"]
0.2
10
0.28
0.2
false
0
false
false
-1
false
true
true
true
0
1
0
low_var_kl
token-mean
0.000001
10
0
0
0.5
1100
constant
0.1
1
22528
32
0
false
fsdp
8
true
false
true
true
false
true
torch
16
0
/root/Qwen2.5-32B
Finished
-
llychinalz
1s
-
["model","optimizer","extra"]
0.2
10
0.28
0.2
false
0
false
false
-1
false
true
true
true
0
1
0
low_var_kl
token-mean
0.000001
10
0
0
0.5
1100
constant
0.1
1
22528
32
0
false
fsdp
8
true
false
true
true
false
true
torch
16
0
/workspace/Qwen2.5-32B
Finished
-
llychinalz
1s
-
["model","optimizer","extra"]
0.2
10
0.28
0.2
false
0
false
false
-1
false
true
true
true
0
1
0
low_var_kl
token-mean
0.000001
10
0
0
0.5
1100
constant
0.1
1
22528
32
0
false
fsdp
8
true
false
true
true
false
true
torch
16
0
/workspace/Qwen2.5-32B
1-7
of 7