重要前提
安装AI Skills的关键前提是:必须科学上网,且开启TUN模式,这一点至关重要,直接决定安装能否顺利完成,在此郑重提醒三遍:科学上网,科学上网,科学上网。查看完整安装教程 →
axolotl by orchestra-research/ai-research-skills
npx skills add https://github.com/orchestra-research/ai-research-skills --skill axolotl提供全面的 axolotl 开发协助,基于官方文档生成。
在以下情况下应触发此技能:
模式 1: 要验证训练任务是否存在可接受的数据传输速度,运行 NCCL 测试可以帮助定位瓶颈,例如:
./build/all_reduce_perf -b 8 -e 128M -f 2 -g 3
模式 2: 在 Axolotl yaml 中配置模型以使用 FSDP。例如:
fsdp_version: 2
fsdp_config:
offload_params: true
state_dict_type: FULL_STATE_DICT
auto_wrap_policy: TRANSFORMER_BASED_WRAP
transformer_layer_cls_to_wrap: LlamaDecoderLayer
reshard_after_forward: true
模式 3: context_parallel_size 应该是 GPU 总数的除数。例如:
context_parallel_size
模式 4: 例如:
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
context_parallel_size=4
模式 5: 在配置中设置 save_compressed: true 可以启用以压缩格式保存模型,这:
save_compressed: true
模式 6: 注意:不必将集成放在 integrations 文件夹中。它可以位于任何位置,只要安装在 Python 环境中的包中即可。请参阅此仓库作为示例:https://github.com/axolotl-ai-cloud/diff-transformer
integrations
模式 7: 处理单样本和批量数据。
utils.trainer.drop_long_seq(sample, sequence_len=2048, min_sequence_len=2)
示例 1 (python):
cli.cloud.modal_.ModalCloud(config, app=None)
示例 2 (python):
cli.cloud.modal_.run_cmd(cmd, run_folder, volumes=None)
示例 3 (python):
core.trainers.base.AxolotlTrainer(
*_args,
bench_data_collator=None,
eval_data_collator=None,
dataset_tags=None,
**kwargs,
)
示例 4 (python):
core.trainers.base.AxolotlTrainer.log(logs, start_time=None)
示例 5 (python):
prompt_strategies.input_output.RawInputOutputPrompter()
此技能包含 references/ 目录中的全面文档:
需要详细信息时,请使用 view 命令读取特定的参考文件。
从 getting_started 或 tutorials 参考文件开始,了解基础概念。
使用相应类别的参考文件(api、guides 等)获取详细信息。
上面的快速参考部分包含从官方文档中提取的常用模式。
从官方来源提取的组织化文档。这些文件包含:
在此处添加用于常见自动化任务的辅助脚本。
在此处添加模板、样板文件或示例项目。
要使用更新的文档刷新此技能:
每周安装次数
63
仓库
GitHub 星标数
5.5K
首次出现
2026年2月7日
安全审计
安装于
codex54
cursor54
opencode54
gemini-cli53
claude-code52
github-copilot52
Comprehensive assistance with axolotl development, generated from official documentation.
This skill should be triggered when:
Pattern 1: To validate that acceptable data transfer speeds exist for your training job, running NCCL Tests can help pinpoint bottlenecks, for example:
./build/all_reduce_perf -b 8 -e 128M -f 2 -g 3
Pattern 2: Configure your model to use FSDP in the Axolotl yaml. For example:
fsdp_version: 2
fsdp_config:
offload_params: true
state_dict_type: FULL_STATE_DICT
auto_wrap_policy: TRANSFORMER_BASED_WRAP
transformer_layer_cls_to_wrap: LlamaDecoderLayer
reshard_after_forward: true
Pattern 3: The context_parallel_size should be a divisor of the total number of GPUs. For example:
context_parallel_size
Pattern 4: For example: - With 8 GPUs and no sequence parallelism: 8 different batches processed per step - With 8 GPUs and context_parallel_size=4: Only 2 different batches processed per step (each split across 4 GPUs) - If your per-GPU micro_batch_size is 2, the global batch size decreases from 16 to 4
context_parallel_size=4
Pattern 5: Setting save_compressed: true in your configuration enables saving models in a compressed format, which: - Reduces disk space usage by approximately 40% - Maintains compatibility with vLLM for accelerated inference - Maintains compatibility with llmcompressor for further optimization (example: quantization)
save_compressed: true
Pattern 6: Note It is not necessary to place your integration in the integrations folder. It can be in any location, so long as it’s installed in a package in your python env. See this repo for an example: https://github.com/axolotl-ai-cloud/diff-transformer
integrations
Pattern 7: Handle both single-example and batched data. - single example: sample[‘input_ids’] is a list[int] - batched data: sample[‘input_ids’] is a list[list[int]]
utils.trainer.drop_long_seq(sample, sequence_len=2048, min_sequence_len=2)
Example 1 (python):
cli.cloud.modal_.ModalCloud(config, app=None)
Example 2 (python):
cli.cloud.modal_.run_cmd(cmd, run_folder, volumes=None)
Example 3 (python):
core.trainers.base.AxolotlTrainer(
*_args,
bench_data_collator=None,
eval_data_collator=None,
dataset_tags=None,
**kwargs,
)
Example 4 (python):
core.trainers.base.AxolotlTrainer.log(logs, start_time=None)
Example 5 (python):
prompt_strategies.input_output.RawInputOutputPrompter()
This skill includes comprehensive documentation in references/:
Use view to read specific reference files when detailed information is needed.
Start with the getting_started or tutorials reference files for foundational concepts.
Use the appropriate category reference file (api, guides, etc.) for detailed information.
The quick reference section above contains common patterns extracted from the official docs.
Organized documentation extracted from official sources. These files contain:
Add helper scripts here for common automation tasks.
Add templates, boilerplate, or example projects here.
To refresh this skill with updated documentation:
Weekly Installs
63
Repository
GitHub Stars
5.5K
First Seen
Feb 7, 2026
Security Audits
Gen Agent Trust HubPassSocketPassSnykWarn
Installed on
codex54
cursor54
opencode54
gemini-cli53
claude-code52
github-copilot52
React 组合模式指南:Vercel 组件架构最佳实践,提升代码可维护性
123,700 周安装
UltraThink Orchestrator - 已弃用的AI工作流编排工具,推荐使用dev-orchestrator替代
87 周安装
dotnet-install:自动化 .NET SDK 和运行时安装指南 - 支持 Windows/macOS/Linux
87 周安装
PydanticAI 模型集成指南:多提供商AI模型调用、备用模型与流式响应
89 周安装
project-discover-preflight-scope:代码库逆向工程范围止损与证据入口盘点方法
87 周安装
Cloudflare Workers 安全指南:10大防护策略与身份验证模式详解
87 周安装
CalDAV日历命令行管理:vdirsyncer+khal同步与查询日历事件教程
87 周安装