Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
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
18 changes: 9 additions & 9 deletions .claude/skills/add-dynamic-filter/SKILL.md
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
---
name: add-dynamic-filter
description: Guide for adding dynamic/filter hooks in slime rollout pipeline. Use when user wants sample-group selection during rollout, buffer filtering before training, or per-sample masking/processing hooks.
description: Guide for adding dynamic/filter hooks in vime rollout pipeline. Use when user wants sample-group selection during rollout, buffer filtering before training, or per-sample masking/processing hooks.
---

# Add Dynamic Filter
Expand All @@ -27,7 +27,7 @@ Use this skill when:

### Step 2: Implement the Function Signature

Dynamic sampling filter (called in `slime/rollout/vllm_rollout.py`):
Dynamic sampling filter (called in `vime/rollout/vllm_rollout.py`):

```python
def filter_function(args, samples, **kwargs):
Expand All @@ -37,12 +37,12 @@ def filter_function(args, samples, **kwargs):
Preferred return type:

```python
from slime.rollout.filter_hub.base_types import DynamicFilterOutput
from vime.rollout.filter_hub.base_types import DynamicFilterOutput

return DynamicFilterOutput(keep=True, reason=None)
```

Buffer filter (called in `slime/rollout/data_source.py`):
Buffer filter (called in `vime/rollout/data_source.py`):

```python
def buffer_filter(args, rollout_id, buffer, num_samples):
Expand Down Expand Up @@ -74,7 +74,7 @@ def process_all_samples(args, all_samples, data_source):
Example wiring:

```bash
--dynamic-sampling-filter-path slime.rollout.filter_hub.dynamic_sampling_filters.check_reward_nonzero_std
--dynamic-sampling-filter-path vime.rollout.filter_hub.dynamic_sampling_filters.check_reward_nonzero_std
--buffer-filter-path <module>.buffer_filter
--rollout-sample-filter-path <module>.rollout_sample_filter
--rollout-all-samples-process-path <module>.process_all_samples
Expand All @@ -95,7 +95,7 @@ Example wiring:

## Reference Locations

- Dynamic filter types: `slime/rollout/filter_hub/base_types.py`
- Dynamic filter example: `slime/rollout/filter_hub/dynamic_sampling_filters.py`
- Rollout generation hook points: `slime/rollout/vllm_rollout.py`
- Buffer filter hook point: `slime/rollout/data_source.py`
- Dynamic filter types: `vime/rollout/filter_hub/base_types.py`
- Dynamic filter example: `vime/rollout/filter_hub/dynamic_sampling_filters.py`
- Rollout generation hook points: `vime/rollout/vllm_rollout.py`
- Buffer filter hook point: `vime/rollout/data_source.py`
12 changes: 6 additions & 6 deletions .claude/skills/add-eval-dataset-config/SKILL.md
Original file line number Diff line number Diff line change
@@ -1,11 +1,11 @@
---
name: add-eval-dataset-config
description: Guide for adding and validating evaluation dataset configuration in slime. Use when user wants to configure eval datasets via --eval-config or --eval-prompt-data, add per-dataset overrides, or customize evaluation rollout behavior.
description: Guide for adding and validating evaluation dataset configuration in vime. Use when user wants to configure eval datasets via --eval-config or --eval-prompt-data, add per-dataset overrides, or customize evaluation rollout behavior.
---

# Add Eval Dataset Config

Configure evaluation datasets in slime with explicit dataset-level overrides and predictable runtime behavior.
Configure evaluation datasets in vime with explicit dataset-level overrides and predictable runtime behavior.

## When to Use

Expand Down Expand Up @@ -53,7 +53,7 @@ eval:

### Step 3: Understand Override Priority

`slime/utils/eval_config.py` resolves fields in this order:
`vime/utils/eval_config.py` resolves fields in this order:

1. Dataset-level values in `eval.datasets[*]`
2. `eval.defaults`
Expand Down Expand Up @@ -85,7 +85,7 @@ Use a separate eval function when inference/eval behavior must differ from train

## Reference Locations

- Eval config model: `slime/utils/eval_config.py`
- Eval config resolution: `slime/utils/arguments.py`
- Eval rollout path: `slime/rollout/vllm_rollout.py`
- Eval config model: `vime/utils/eval_config.py`
- Eval config resolution: `vime/utils/arguments.py`
- Eval rollout path: `vime/rollout/vllm_rollout.py`
- Customization docs: `docs/en/get_started/customization.md`
16 changes: 8 additions & 8 deletions .claude/skills/add-reward-function/SKILL.md
Original file line number Diff line number Diff line change
@@ -1,11 +1,11 @@
---
name: add-reward-function
description: Guide for adding a custom reward function in slime and wiring it through --custom-rm-path (and optional reward post-processing). Use when user wants new reward logic, remote/service reward integration, or task-specific reward shaping.
description: Guide for adding a custom reward function in vime and wiring it through --custom-rm-path (and optional reward post-processing). Use when user wants new reward logic, remote/service reward integration, or task-specific reward shaping.
---

# Add Reward Function

Implement custom reward logic and connect it to slime rollout/training safely.
Implement custom reward logic and connect it to vime rollout/training safely.

## When to Use

Expand All @@ -24,11 +24,11 @@ Pick one of these:
- Single-sample mode (`--group-rm` disabled): custom function gets one `Sample`
- Group/batch mode (`--group-rm` enabled): custom function gets `list[Sample]`

`slime.rollout.rm_hub.__init__.py` calls your function via `--custom-rm-path`.
`vime.rollout.rm_hub.__init__.py` calls your function via `--custom-rm-path`.

### Step 2: Create Reward Module

Create `slime/rollout/rm_hub/<your_rm>.py`.
Create `vime/rollout/rm_hub/<your_rm>.py`.

Supported signatures:

Expand Down Expand Up @@ -66,14 +66,14 @@ Wire with:
--custom-reward-post-process-path <module>.post_process_rewards
```

This hook is consumed in `slime/ray/rollout.py`.
This hook is consumed in `vime/ray/rollout.py`.

### Step 5: Wire and Validate

Use:

```bash
--custom-rm-path slime.rollout.rm_hub.<your_rm>.custom_rm
--custom-rm-path vime.rollout.rm_hub.<your_rm>.custom_rm
```

## Common Mistakes
Expand All @@ -85,6 +85,6 @@ Use:

## Reference Locations

- Reward dispatch: `slime/rollout/rm_hub/__init__.py`
- Reward post-process hook: `slime/ray/rollout.py`
- Reward dispatch: `vime/rollout/rm_hub/__init__.py`
- Reward post-process hook: `vime/ray/rollout.py`
- Customization docs: `docs/en/get_started/customization.md`
30 changes: 15 additions & 15 deletions .claude/skills/add-rollout-function/SKILL.md
Original file line number Diff line number Diff line change
@@ -1,18 +1,18 @@
---
name: add-rollout-function
description: Guide for adding a new rollout function in slime and wiring it through --rollout-function-path. Use when user wants to implement custom rollout data generation logic, custom train/eval rollout outputs, or migrate from the default vLLM rollout path.
description: Guide for adding a new rollout function in vime and wiring it through --rollout-function-path. Use when user wants to implement custom rollout data generation logic, custom train/eval rollout outputs, or migrate from the default vLLM rollout path.
---

# Add Rollout Function

Implement a custom rollout function and integrate it safely with slime training/eval flow.
Implement a custom rollout function and integrate it safely with vime training/eval flow.

## When to Use

Use this skill when:

- User asks to add a new rollout task or rollout generation function
- User asks to replace default `slime.rollout.vllm_rollout.generate_rollout`
- User asks to replace default `vime.rollout.vllm_rollout.generate_rollout`
- User asks to customize train/eval data generation behavior

## Step-by-Step Guide
Expand All @@ -21,15 +21,15 @@ Use this skill when:

Start from one of these references:

- Async RL-style rollout: `slime/rollout/vllm_rollout.py`
- Simple SFT-style rollout: `slime/rollout/sft_rollout.py`
- Async RL-style rollout: `vime/rollout/vllm_rollout.py`
- Simple SFT-style rollout: `vime/rollout/sft_rollout.py`

If the task needs engine-based async generation and rewards, use the vLLM path as base.
If the task is file/buffer-driven and simple, use sft path as base.

### Step 2: Create the New Rollout Module

Create a new file, for example: `slime/rollout/<your_rollout>.py`
Create a new file, for example: `vime/rollout/<your_rollout>.py`

Required callable signature:

Expand All @@ -38,7 +38,7 @@ def generate_rollout(args, rollout_id, data_source, evaluation=False) -> Rollout
...
```

Return types are defined in `slime/rollout/base_types.py`.
Return types are defined in `vime/rollout/base_types.py`.

### Step 3: Implement Train and Eval Branches Explicitly

Expand All @@ -48,7 +48,7 @@ Return types are defined in `slime/rollout/base_types.py`.
Minimal skeleton:

```python
from slime.rollout.base_types import RolloutFnTrainOutput, RolloutFnEvalOutput
from vime.rollout.base_types import RolloutFnTrainOutput, RolloutFnEvalOutput


def generate_rollout(args, rollout_id, data_source, evaluation=False):
Expand Down Expand Up @@ -83,12 +83,12 @@ If partial rollout or masking logic is involved, keep `loss_mask` semantics cons
Set your function path via CLI:

```bash
--rollout-function-path slime.rollout.<your_rollout>.generate_rollout
--rollout-function-path vime.rollout.<your_rollout>.generate_rollout
```

The default and signature expectation are documented in:

- `slime/utils/arguments.py`
- `vime/utils/arguments.py`
- `docs/en/get_started/customization.md`

## Common Mistakes
Expand All @@ -100,9 +100,9 @@ The default and signature expectation are documented in:

## Reference Locations

- Default rollout: `slime/rollout/vllm_rollout.py`
- Simple custom example: `slime/rollout/sft_rollout.py`
- Output dataclasses: `slime/rollout/base_types.py`
- Wiring/loading: `slime/ray/rollout.py`
- Argument definition: `slime/utils/arguments.py`
- Default rollout: `vime/rollout/vllm_rollout.py`
- Simple custom example: `vime/rollout/sft_rollout.py`
- Output dataclasses: `vime/rollout/base_types.py`
- Wiring/loading: `vime/ray/rollout.py`
- Argument definition: `vime/utils/arguments.py`
- Customization docs: `docs/en/get_started/customization.md`
4 changes: 2 additions & 2 deletions .claude/skills/add-tests-and-ci/SKILL.md
Original file line number Diff line number Diff line change
@@ -1,11 +1,11 @@
---
name: add-tests-and-ci
description: Guide for adding or updating slime tests and CI wiring. Use when tasks require new test cases, CI registration, test matrix updates, or workflow template changes.
description: Guide for adding or updating vime tests and CI wiring. Use when tasks require new test cases, CI registration, test matrix updates, or workflow template changes.
---

# Add Tests and CI

Add reliable tests and integrate them with slime CI flow.
Add reliable tests and integrate them with vime CI flow.

## When to Use

Expand Down
22 changes: 11 additions & 11 deletions .github/ISSUE_TEMPLATE/bug_report.yml
Original file line number Diff line number Diff line change
@@ -1,5 +1,5 @@
name: Bug Report
description: Report a bug to help us improve slime. Please check the docs and existing issues before submitting.
description: Report a bug to help us improve vime. Please check the docs and existing issues before submitting.
title: "[Bug] "
labels: ["bug"]
body:
Expand All @@ -8,13 +8,13 @@ body:
value: |
Thank you for taking the time to report a bug!

> **Note:** slime's open-source collaboration is currently focused on **bug fixes** and **general-purpose RL optimizations**.
> Feature requests are outside our current scope — please see [CONTRIBUTING.md](https://github.com/THUDM/slime/blob/main/CONTRIBUTING.md) for details.
> **Note:** vime's open-source collaboration is currently focused on **bug fixes** and **general-purpose RL optimizations**.
> Feature requests are outside our current scope — please see [CONTRIBUTING.md](https://github.com/vllm-project/vime/blob/main/CONTRIBUTING.md) for details.

Before submitting, please make sure you have:
- [ ] Read the [documentation](https://thudm.github.io/slime/) and [FAQ](https://thudm.github.io/slime/en/get_started/qa.html)
- [ ] Searched [existing issues](https://github.com/THUDM/slime/issues) to avoid duplicates
- [ ] Reproduced the issue with the latest version of slime
- [ ] Read the [documentation](https://vllm-project.github.io/vime/) and [FAQ](https://vllm-project.github.io/vime/en/get_started/qa.html)
- [ ] Searched [existing issues](https://github.com/vllm-project/vime/issues) to avoid duplicates
- [ ] Reproduced the issue with the latest version of vime

- type: textarea
id: description
Expand Down Expand Up @@ -60,9 +60,9 @@ body:
attributes:
label: Environment
description: |
Please provide your environment details. You can run `pip show slime` and `python -c "import torch; print(torch.__version__)"` to get some of this info.
Please provide your environment details. You can run `pip show vime` and `python -c "import torch; print(torch.__version__)"` to get some of this info.
value: |
- slime version:
- vime version:
- Python version:
- PyTorch version:
- CUDA version:
Expand Down Expand Up @@ -93,11 +93,11 @@ body:
attributes:
label: Pre-submission Checklist
options:
- label: I have read the [CONTRIBUTING.md](https://github.com/THUDM/slime/blob/main/CONTRIBUTING.md) and understand the collaboration scope.
- label: I have read the [CONTRIBUTING.md](https://github.com/vllm-project/vime/blob/main/CONTRIBUTING.md) and understand the collaboration scope.
required: true
- label: I have read the [documentation](https://thudm.github.io/slime/) and my issue is not addressed there.
- label: I have read the [documentation](https://vllm-project.github.io/vime/) and my issue is not addressed there.
required: true
- label: I have searched for [existing issues](https://github.com/THUDM/slime/issues) and this is not a duplicate.
- label: I have searched for [existing issues](https://github.com/vllm-project/vime/issues) and this is not a duplicate.
required: true
- label: I have provided a minimal, reproducible example.
required: true
8 changes: 4 additions & 4 deletions .github/ISSUE_TEMPLATE/config.yml
Original file line number Diff line number Diff line change
@@ -1,16 +1,16 @@
blank_issues_enabled: false
contact_links:
- name: Collaboration Scope / 协作范围
url: https://github.com/THUDM/slime/blob/main/CONTRIBUTING.md
url: https://github.com/vllm-project/vime/blob/main/CONTRIBUTING.md
about: |
Please read before submitting. We welcome bug reports and general RL optimizations; feature requests are currently outside our scope.
提交前请阅读。我们欢迎 bug 报告和通用 RL 优化,feature request 暂时不在协作范围内。
- name: Documentation
url: https://thudm.github.io/slime/
url: https://vllm-project.github.io/vime/
about: Please check the documentation before opening an issue. Many common questions are already answered there.
- name: Q&A / FAQ
url: https://thudm.github.io/slime/en/get_started/qa.html
url: https://vllm-project.github.io/vime/en/get_started/qa.html
about: Check the FAQ — your question might already have an answer.
- name: Discussions
url: https://github.com/THUDM/slime/discussions
url: https://github.com/vllm-project/vime/discussions
about: For general questions and open-ended conversations, please use GitHub Discussions instead.
24 changes: 12 additions & 12 deletions .github/ISSUE_TEMPLATE/question.yml
Original file line number Diff line number Diff line change
@@ -1,5 +1,5 @@
name: Question / Help
description: Ask a question about using slime. Please check the docs and FAQ first.
description: Ask a question about using vime. Please check the docs and FAQ first.
title: "[Question] "
labels: ["question"]
body:
Expand All @@ -8,16 +8,16 @@ body:
value: |
Thank you for reaching out!

> **Note:** slime's open-source collaboration is currently focused on **bug fixes** and **general-purpose RL optimizations**.
> Feature requests are outside our current scope — please see [CONTRIBUTING.md](https://github.com/THUDM/slime/blob/main/CONTRIBUTING.md) for details.
> **Note:** vime's open-source collaboration is currently focused on **bug fixes** and **general-purpose RL optimizations**.
> Feature requests are outside our current scope — please see [CONTRIBUTING.md](https://github.com/vllm-project/vime/blob/main/CONTRIBUTING.md) for details.

Before asking, please make sure you have checked:
- [Quick Start Guide](https://thudm.github.io/slime/en/get_started/quick_start.html)
- [Usage Guide](https://thudm.github.io/slime/en/get_started/usage.html)
- [FAQ](https://thudm.github.io/slime/en/get_started/qa.html)
- [Examples](https://github.com/THUDM/slime/tree/main/examples)
- [Quick Start Guide](https://vllm-project.github.io/vime/en/get_started/quick_start.html)
- [Usage Guide](https://vllm-project.github.io/vime/en/get_started/usage.html)
- [FAQ](https://vllm-project.github.io/vime/en/get_started/qa.html)
- [Examples](https://github.com/vllm-project/vime/tree/main/examples)

Many common questions are already covered in the documentation above. If your question is about general usage or open-ended discussion, consider using [GitHub Discussions](https://github.com/THUDM/slime/discussions) instead.
Many common questions are already covered in the documentation above. If your question is about general usage or open-ended discussion, consider using [GitHub Discussions](https://github.com/vllm-project/vime/discussions) instead.

- type: textarea
id: question
Expand Down Expand Up @@ -46,7 +46,7 @@ body:
label: Environment (if relevant)
description: If your question is related to setup or runtime behavior, please provide your environment details.
value: |
- slime version:
- vime version:
- Python version:
- PyTorch version:
- CUDA version:
Expand All @@ -64,9 +64,9 @@ body:
attributes:
label: Pre-submission Checklist
options:
- label: I have read the [CONTRIBUTING.md](https://github.com/THUDM/slime/blob/main/CONTRIBUTING.md) and understand the collaboration scope.
- label: I have read the [CONTRIBUTING.md](https://github.com/vllm-project/vime/blob/main/CONTRIBUTING.md) and understand the collaboration scope.
required: true
- label: I have read the [documentation](https://thudm.github.io/slime/) and [FAQ](https://thudm.github.io/slime/en/get_started/qa.html) and my question is not answered there.
- label: I have read the [documentation](https://vllm-project.github.io/vime/) and [FAQ](https://vllm-project.github.io/vime/en/get_started/qa.html) and my question is not answered there.
required: true
- label: I have searched for [existing issues](https://github.com/THUDM/slime/issues) and my question has not been asked before.
- label: I have searched for [existing issues](https://github.com/vllm-project/vime/issues) and my question has not been asked before.
required: true
Loading
Loading