Skip to content

chore: rename deprecated orchestrator config keys #2327

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
mikasenghaas merged 1 commit into from
Apr 19, 2026
Merged

chore: rename deprecated orchestrator config keys #2327

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions configs/acereason_math/stage1.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,11 +19,11 @@ name = "stage1"
batch_size = 1024
rollouts_per_example = 8

[orchestrator.sampling]
[orchestrator.train.sampling]
temperature = 0.6
max_tokens = 8192
max_completion_tokens = 8192

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "math-env"
args = { dataset_name = "nvidia/AceReason-Math", dataset_subset = "default", question_key = "problem", math_verify_max_workers = 128, math_verify_timeout = 60 }

Expand Down
6 changes: 3 additions & 3 deletions configs/acereason_math/stage2.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,11 +20,11 @@ name = "stage2"
batch_size = 2048
rollouts_per_example = 16

[orchestrator.sampling]
[orchestrator.train.sampling]
temperature = 0.6
max_tokens = 16384
max_completion_tokens = 16384

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "math-env"
args = { dataset_name = "nvidia/AceReason-Math", dataset_subset = "default", question_key = "problem", math_verify_max_workers = 128, math_verify_timeout = 60 }

Expand Down
6 changes: 3 additions & 3 deletions configs/alphabet_sort/rl.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,10 +13,10 @@ name = "Qwen/Qwen3-4B-Instruct-2507"
batch_size = 512
rollouts_per_example = 16

[orchestrator.sampling]
max_tokens = 512
[orchestrator.train.sampling]
max_completion_tokens = 512

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "alphabet-sort"
name = "alphabet-sort"
args = { min_turns = 2, max_turns = 2 }
Expand Down
6 changes: 3 additions & 3 deletions configs/ci/integration/alphabet_sort/start.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,10 +13,10 @@ lr = 1e-5
batch_size = 128
rollouts_per_example = 8

[orchestrator.sampling]
max_tokens = 384
[orchestrator.train.sampling]
max_completion_tokens = 384

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "alphabet-sort"
name = "alphabet-sort"
args = { min_turns = 2, max_turns = 2, min_names_per_turn = 1, max_names_per_turn = 3, similarity_power = 4, power_per_turn = false }
Expand Down
6 changes: 3 additions & 3 deletions configs/ci/integration/alphabet_sort_branch/start.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,10 +18,10 @@ lr = 1e-5
batch_size = 128
rollouts_per_example = 8

[orchestrator.sampling]
max_tokens = 1024
[orchestrator.train.sampling]
max_completion_tokens = 1024

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "primeintellect/alphabet-sort"
name = "alphabet-sort"
args = { min_turns = 2, max_turns = 2, min_names_per_turn = 1, max_names_per_turn = 3, similarity_power = 4, power_per_turn = false }
Expand Down
6 changes: 3 additions & 3 deletions configs/ci/integration/rl/resume.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,10 +14,10 @@ lr = 3e-6
batch_size = 128
rollouts_per_example = 16

[orchestrator.sampling]
max_tokens = 128
[orchestrator.train.sampling]
max_completion_tokens = 128

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "reverse-text"

[inference]
6 changes: 3 additions & 3 deletions configs/ci/integration/rl/start.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,10 +13,10 @@ lr = 3e-6
batch_size = 128
rollouts_per_example = 16

[orchestrator.sampling]
max_tokens = 128
[orchestrator.train.sampling]
max_completion_tokens = 128

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "reverse-text"

[inference]
6 changes: 3 additions & 3 deletions configs/ci/integration/rl_lora/resume.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,10 +23,10 @@ rollouts_per_example = 16
[orchestrator.model.lora]
name = "r8-1e-4"

[orchestrator.sampling]
max_tokens = 128
[orchestrator.train.sampling]
max_completion_tokens = 128

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "reverse-text"

[inference]
Expand Down
6 changes: 3 additions & 3 deletions configs/ci/integration/rl_lora/start.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,10 +22,10 @@ rollouts_per_example = 16
[orchestrator.model.lora]
name = "r8-1e-4"

[orchestrator.sampling]
max_tokens = 128
[orchestrator.train.sampling]
max_completion_tokens = 128

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "reverse-text"

[inference]
Expand Down
6 changes: 3 additions & 3 deletions configs/ci/integration/rl_moe/start.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,10 +13,10 @@ lr = 3e-6
batch_size = 128
rollouts_per_example = 16

[orchestrator.sampling]
max_tokens = 128
[orchestrator.train.sampling]
max_completion_tokens = 128

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "reverse-text"

[inference]
Expand Down
6 changes: 3 additions & 3 deletions configs/ci/integration/rl_multi_run/orchestrator.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,10 +11,10 @@ name = "PrimeIntellect/Qwen3-0.6B-Reverse-Text-SFT"
[optim]
lr = 3e-5

[sampling]
max_tokens = 128
[train.sampling]
max_completion_tokens = 128

[[env]]
[[train.env]]
id = "reverse-text"

[ckpt]
Expand Down
6 changes: 3 additions & 3 deletions configs/ci/nightly/acereason_math.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,11 +17,11 @@ name = "deepseek-ai/DeepSeek-R1-Distill-Qwen-7B"
batch_size = 1024
rollouts_per_example = 8

[orchestrator.sampling]
[orchestrator.train.sampling]
temperature = 0.6
max_tokens = 8192
max_completion_tokens = 8192

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "math-env"
name = "acereason-math"
args = { dataset_name = "nvidia/AceReason-Math", dataset_subset = "default", question_key = "problem", math_verify_max_workers = 128, math_verify_timeout = 60 }
Expand Down
6 changes: 3 additions & 3 deletions configs/ci/nightly/multimodal_color_codeword.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,10 +17,10 @@ language_model_attr = "model.language_model"
batch_size = 256
rollouts_per_example = 16

[orchestrator.sampling]
max_tokens = 64
[orchestrator.train.sampling]
max_completion_tokens = 64

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "color-codeword"
args = { images_per_turn = 1, max_turns = 3, num_examples = 1000, seed = 42 }

Expand Down
4 changes: 2 additions & 2 deletions configs/debug/orch.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,6 @@ max_steps = 5
max_async_level = 5
batch_size = 16

[sampling]
max_tokens = 16
[train.sampling]
max_completion_tokens = 16

6 changes: 3 additions & 3 deletions configs/deepscaler/stage1.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,11 +19,11 @@ interval = 100
batch_size = 1024
rollouts_per_example = 8

[orchestrator.sampling]
[orchestrator.train.sampling]
temperature = 0.6
max_tokens = 8192
max_completion_tokens = 8192

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "math-env"
name = "deepscaler"
args = { dataset_name= "agentica-org/DeepScaleR-Preview-Dataset", dataset_subset = "default", question_key = "problem", answer_key = "solution", math_verify_max_workers = 128, math_verify_timeout = 60 }
Expand Down
6 changes: 3 additions & 3 deletions configs/deepscaler/stage2.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,11 +20,11 @@ resume_step = 500
batch_size = 1024
rollouts_per_example = 8

[orchestrator.sampling]
[orchestrator.train.sampling]
temperature = 0.6
max_tokens = 16384
max_completion_tokens = 16384

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "math-env"
name = "deepscaler"
args = { dataset_name= "agentica-org/DeepScaleR-Preview-Dataset", dataset_subset = "default", question_key = "problem", answer_key = "solution", math_verify_max_workers = 128, math_verify_timeout = 60 }
Expand Down
6 changes: 3 additions & 3 deletions configs/deepscaler/stage3.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,11 +20,11 @@ resume_step = 1000
batch_size = 1024
rollouts_per_example = 8

[orchestrator.sampling]
[orchestrator.train.sampling]
temperature = 0.6
max_tokens = 24576
max_completion_tokens = 24576

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "math-env"
name = "deepscaler"
args = { dataset_name= "agentica-org/DeepScaleR-Preview-Dataset", dataset_subset = "default", question_key = "problem", answer_key = "solution", math_verify_max_workers = 128, math_verify_timeout = 60 }
Expand Down
6 changes: 3 additions & 3 deletions configs/elastic/rl.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -32,15 +32,15 @@ lr = 1e-5
batch_size = 512
rollouts_per_example = 8

[orchestrator.sampling]
max_tokens = 768
[orchestrator.train.sampling]
max_completion_tokens = 768

[orchestrator.client.elastic]
hostname = "localhost"
port = 8000
sync_interval = 5.0

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "alphabet-sort"
name = "alphabet-sort"
args = { min_turns = 3, max_turns = 5, power_per_turn = false }
10 changes: 5 additions & 5 deletions configs/env_mix/env_mix.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,27 +16,27 @@ name = "Qwen/Qwen3-4B-Instruct-2507"
batch_size = 512
rollouts_per_example = 16

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "math-env"
name = "math"
args = { min_avg_reward = 0.1, rubric_max_workers = 128 }

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "code-env"
name = "code"
args = { min_avg_reward = 0.1 }

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "science-env"
name = "science"
args = { min_avg_reward = 0.1, pool_size = 128, rubric_max_workers = 128 }

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "logic-env"
name = "logic"
args = { min_avg_reward = 0.1 }

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "math-env"
name = "math-python"
args = { min_avg_reward = 0.1, sandbox_client_max_workers = 128, sandbox_timeout_minutes = 10, sandbox_memory_gb = 1, sandbox_disk_size_gb = 1, pip_install_packages = "numpy sympy", python_tool = true, rubric_max_workers = 128 }
Expand Down
6 changes: 3 additions & 3 deletions configs/gsm8k/rl.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,10 +12,10 @@ name = "PrimeIntellect/Qwen3-0.6B"
batch_size = 512
rollouts_per_example = 16

[orchestrator.sampling]
max_tokens = 2048
[orchestrator.train.sampling]
max_completion_tokens = 2048

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "math-env"
name = "gsm8k"
args = { dataset_name = "openai/gsm8k", dataset_subset = "main", math_verify_max_workers = 128, math_verify_timeout = 60 }
Expand Down
8 changes: 4 additions & 4 deletions configs/hendrycks_math/rl.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,10 +12,10 @@ name = "Qwen/Qwen3-4B-Instruct-2507"
batch_size = 512
rollouts_per_example = 16

[orchestrator.sampling]
max_tokens = 2048
[orchestrator.train.sampling]
max_completion_tokens = 2048

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "math-env"
name = "hendrycks-math"
args = { dataset_name = "PrimeIntellect/Hendrycks-Math", dataset_subset = "default", math_verify_max_workers = 128, math_verify_timeout = 60 }
Expand All @@ -28,7 +28,7 @@ hard_threshold = 0.0
interval = 10

[orchestrator.eval.sampling]
max_tokens = 2048
max_completion_tokens = 2048

[[orchestrator.eval.env]]
id = "math500"
Expand Down
2 changes: 1 addition & 1 deletion configs/hendrycks_math/sanity.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,7 +12,7 @@ name = "deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B"
batch_size = 512
rollouts_per_example = 8

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "math-env"
args = { dataset_name = "mikasenghaas/Sanity-Test-R1D-1.5B", dataset_subset = "default" }
name = "hendrycks-math"
Expand Down
4 changes: 2 additions & 2 deletions configs/math_group/rl.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,13 +11,13 @@ name = "Qwen/Qwen3-4B-Instruct-2507"
batch_size = 256
rollouts_per_example = 8

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "math-env"
name = "hendrycks-math"
args = { dataset_name = "PrimeIntellect/Hendrycks-Math", dataset_subset = "default" }
ratio = 0.5

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "math-env"
name = "acereason-math"
args = { dataset_name = "nvidia/AceReason-Math", dataset_subset = "default", question_key = "problem" }
Expand Down
6 changes: 3 additions & 3 deletions configs/math_python/math_python.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,10 +12,10 @@ name = "Qwen/Qwen3-4B-Instruct-2507"
batch_size = 512
rollouts_per_example = 16

[orchestrator.sampling]
max_tokens = 512
[orchestrator.train.sampling]
max_completion_tokens = 512

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "math-python"
args = { sandbox_client_max_workers = 128, sandbox_timeout_minutes = 10, sandbox_memory_gb = 1, sandbox_disk_size_gb = 1, pip_install_packages = "numpy sympy" }

Expand Down
6 changes: 3 additions & 3 deletions configs/multimodal/rl_color_codeword.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,10 +13,10 @@ batch_size = 256
rollouts_per_example = 16


[orchestrator.sampling]
max_tokens = 64
[orchestrator.train.sampling]
max_completion_tokens = 64

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "color-codeword"
args = { images_per_turn = 1, max_turns = 3, num_examples = 1000, seed = 42 }

Expand Down
6 changes: 3 additions & 3 deletions configs/multimodal/rl_color_codeword_test.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,10 +13,10 @@ language_model_attr = "model.language_model"
batch_size = 16
rollouts_per_example = 2

[orchestrator.sampling]
max_tokens = 32
[orchestrator.train.sampling]
max_completion_tokens = 32

[[orchestrator.env]]
[[orchestrator.train.env]]
id = "color-codeword"
args = { images_per_turn = 1, max_turns = 2, num_examples = 100, seed = 42 }

Expand Down
Loading
Loading