Skip to content

[model] support nemotron transcribe #10458

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
hjh0119 merged 9 commits into from
Oct 29, 2026
+555 −16
Merged

[model] support nemotron transcribe #10458

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
245e1d3
init commit
hjh0119 Oct 28, 2026
2ab343f
Merge remote-tracking branch 'origin' into transcribe
hjh0119 Oct 28, 2026
bddb76b
upadte model doc
hjh0119 Oct 28, 2026
9ea65cc
update
hjh0119 Oct 28, 2026
2801d82
update dataset doc
hjh0119 Oct 29, 2026
c5b3499
remove unavailable models
hjh0119 Oct 29, 2026
8e31ca7
compat transformesr
hjh0119 Oct 29, 2026
9721bc2
lint&update news date
hjh0119 Oct 29, 2026
5f45ba4
clean script
hjh0119 Oct 29, 2026
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
7 changes: 4 additions & 3 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -77,6 +77,7 @@ You can contact us and communicate with us by adding our group:


## 🎉 News
- 🔥 2026.10.29: Day-0 Support for [NVIDIA-Nemotron-3.5-Transcribe](nv-community/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B), Refer to the [SFT examples](examples/models/nemotron_transcribe).
- 🔥 2026.10.22: Support [EntroPack](https://github.com/modelscope/entropack) flexible bitrate model quantization, enabling fine-grained control of the model's bitrate based on the VRAM budget, with support for Transformers inference and LoRA training. See the [best practices](docs/source_en/BestPractices/Flexble-Quantization-With-Entropack.md) and the [QLoRA example](examples/train/qlora/entropack.sh).
- 🔥 2026.10.19: Support for [EmbeddingGemma-2](https://www.modelscope.cn/models/google/embeddinggemma-2) (740M, multimodal embedding model, unifying text/image/video/audio in a single 768d vector space). Inference and training verified with infonce loss fine-tuning.
- 🔥 2026.10.15: Day-0 Support for [NVIDIA-Nemotron-3.5-Super-VL-120B-A12B](https://www.modelscope.cn/models/nv-community/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16), Refer to the [SFT examples](examples/models/nemotron_super_vl).
Expand All @@ -90,6 +91,9 @@ You can contact us and communicate with us by adding our group:
- 🎁 2026.07.22: Support inference for the moonshotai multimodal model [Kimi-K3](https://modelscope.cn/models/moonshotai/Kimi-K3), including the XTML chat template, thinking channel (`reasoning_effort`) and tool calling (`--agent_template kimi_k3`).
- 🎁 2026.06.10: Megatron-Ray now supports GRPO and GKD training. See [docs](./docs/source_en/Instruction/Ray.md) and [examples](examples/ray).
- 🎁 2026.03.03: **ms-swift v4.0** major version is officially released. For release notes, please refer to [here](https://github.com/modelscope/ms-swift/releases/tag/v4.0.0). You can provide your suggestions to us in [this issue](https://github.com/modelscope/ms-swift/issues/7250). Thank you for your support.


<details><summary>More</summary>
- 🎁 2025.11.14: Megatron GRPO is now available! Check out the [docs](./docs/source_en/Megatron-SWIFT/GRPO.md) and [examples](examples/megatron/grpo).
- 🎁 2025.11.04: Support for [Mcore-Bridge](docs/source_en/Megatron-SWIFT/Mcore-Bridge.md), making Megatron training as simple and easy to use as transformers.
- 🎁 2025.10.28: Ray [here](docs/source_en/Instruction/Ray.md).
Expand All @@ -100,9 +104,6 @@ You can contact us and communicate with us by adding our group:
- 🎁 2025.07.09: Megatron-SWIFT supports LoRA training. Compared to ms-swift, it achieves significant speedup on MoE models. Training scripts can be found [here](https://github.com/modelscope/ms-swift/blob/main/examples/megatron/lora).
- 🎁 2025.06.23: Fine-tuning of reranker models is supported. Training scripts can be found here: [Reranker](https://github.com/modelscope/ms-swift/blob/main/examples/train/reranker/train_reranker.sh).
- 🎁 2025.06.15: Support for GKD training on both pure text large models and multimodal models. Training scripts can be found here: [Pure Text](https://github.com/modelscope/ms-swift/blob/main/examples/train/rlhf/gkd), [Multimodal](https://github.com/modelscope/ms-swift/blob/main/examples/train/multimodal/rlhf/gkd).

<details><summary>More</summary>

- 🎁 2025.06.11: Support for using Megatron parallelism techniques for RLHF training. The training script can be found [here](https://github.com/modelscope/ms-swift/tree/main/examples/megatron/rlhf).
- 🎁 2025.05.29: Support sequence parallel in pretrain, sft, dpo and grpo, check script [here](https://github.com/modelscope/ms-swift/tree/main/examples/train/sequence_parallel).
- 🎁 2025.05.11: GRPO now supports custom processing logic for reward models. See the GenRM example [here](./docs/source_en/Instruction/GRPO/DeveloperGuide/reward_model.md).
Expand Down
9 changes: 5 additions & 4 deletions README_CN.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -73,9 +73,10 @@
- **模型量化**：支持AWQ、GPTQ、FP8和BNB的量化导出，导出的模型支持使用vLLM/SGLang/LmDeploy推理加速。

## 🎉 新闻
- 🔥 2026.10.29: Day-0 支持 [NVIDIA-Nemotron-3.5-Transcribe](nv-community/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B)，训练示例参考 [SFT示例](examples/models/nemotron_transcribe)。
- 🔥 2026.10.22: 支持 [EntroPack](https://github.com/modelscope/entropack) 灵活码率模型量化，可根据显存预算精细控制模型码率，支持 Transformers 推理和 LoRA 训练。参考[最佳实践](docs/source/BestPractices/Flexble-Quantization-With-Entropack.md)和 [QLoRA 示例](examples/train/qlora/entropack.sh)。
- 🔥 2026.10.19: 支持 [EmbeddingGemma-2](https://www.modelscope.cn/models/google/embeddinggemma-2)（740M，多模态嵌入模型，支持文本/图像/视频/音频统一向量空间），支持 infonce loss 微调。
- 🔥 2026.10.15: Day-0 支持 [NVIDIA-Nemotron-3.5-Super-VL-120B-A12B](https://www.modelscope.cn/models/nv-community/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16)，训练示例参考 [SFT示例](examples/models/nemotron_super_vl).
- 🔥 2026.10.15: Day-0 支持 [NVIDIA-Nemotron-3.5-Super-VL-120B-A12B](https://www.modelscope.cn/models/nv-community/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16)，训练示例参考 [SFT示例](examples/models/nemotron_super_vl)。
- 🔥 2026.10.13: 支持Decision模型（类型化决策System-1打分模型）的训练与推理，使用 `--task_type decision`，通过单次前向传播输出每个问题选项集上的校准概率。支持 [JEV-27B-VL](https://www.modelscope.cn/models/autotrust/JEV-27B-VL)、[Clef](https://www.modelscope.cn/models/Cloudflare/clef)、[OmniJev](https://modelscope.cn/models/AI-ModelScope/OmniJev) 和 [Kev](https://www.modelscope.cn/models/jaredpalmer/kev-4b)（0.8b/4b/9b），内置 distill_rps、smooth_brier、ce_rps 等proper scoring rule损失函数。文档参考[这里](docs/source/BestPractices/Decision.md)，训练示例参考[这里](examples/train/decision)。
- 🔥 2026.10.01: 支持 [Xing4.0-29B-A4B](https://www.modelscope.cn/models/XingChen-AGI/Xing4.0-29B-A4B)：transformers 后端的训练与推理，支持单/多卡（DDP 与 DeepSpeed ZeRO-3）的 LoRA 与全参数 SFT，以及 grouped_mm 专家加速（~3.7x），示例参考 [这里](examples/models/xing4_0)。
- 🔥 2026.09.21: 支持 [DeepSeek-V4.1-Flash](https://www.modelscope.cn/models/deepseek-ai/DeepSeek-V4.1-Flash)：Megatron-SWIFT 全参数训练（SFT 与 GRPO）以及 vLLM 推理，示例参考 [这里](examples/models/deepseek_v41)。
Expand All @@ -86,6 +87,9 @@
- 🎁 2026.07.22: 支持moonshotai多模态模型[Kimi-K3](https://modelscope.cn/models/moonshotai/Kimi-K3)的推理，包含XTML对话模板、思考通道（`reasoning_effort`）与工具调用（`--agent_template kimi_k3`）。
- 🎁 2026.06.10: Megatron-Ray支持GRPO和GKD训练，查看[文档](docs/source/Instruction/Ray.md)和[示例](examples/ray)。
- 🎁 2026.03.03: **ms-swift v4.0**大版本正式发布，release note参考[这里](https://github.com/modelscope/ms-swift/releases/tag/v4.0.0)，您的建议可以在[这个issue](https://github.com/modelscope/ms-swift/issues/7250)中反馈给我们，感谢您的支持。


<details><summary>更多</summary>
- 🎁 2025.11.14: Megatron GRPO现已支持！查看[文档](./docs/source/Megatron-SWIFT/GRPO.md)和[示例](examples/megatron/grpo)。
- 🎁 2025.11.04: 支持[Mcore-Bridge](docs/source/Megatron-SWIFT/Mcore-Bridge.md)，使Megatron训练像transformers一样简单易用。
- 🎁 2025.10.28: Ray [已支持](docs/source/Instruction/Ray.md)。
Expand All @@ -96,9 +100,6 @@
- 🎁 2025.07.09: Megatron-SWIFT支持LoRA训练。相比ms-swift，在MoE模型提速显著。训练脚本参考[这里](https://github.com/modelscope/ms-swift/blob/main/examples/megatron/lora)。
- 🎁 2025.06.23: 支持Reranker模型训练，训练脚本参考[这里](https://github.com/modelscope/ms-swift/blob/main/examples/train/reranker/train_reranker.sh)。
- 🎁 2025.06.15: 支持对纯文本大模型和多模态模型进行GKD训练。训练脚本参考这里：[纯文本](https://github.com/modelscope/ms-swift/blob/main/examples/train/rlhf/gkd), [多模态](https://github.com/modelscope/ms-swift/blob/main/examples/train/multimodal/rlhf/gkd)。

<details><summary>更多</summary>

- 🎁 2025.06.11: 支持使用Megatron并行技术进行RLHF训练，训练脚本参考[这里](https://github.com/modelscope/ms-swift/tree/main/examples/megatron/rlhf)。
- 🎁 2025.05.29: 支持pt、sft、dpo、grpo的序列并行，具体请查看[脚本](https://github.com/modelscope/ms-swift/tree/main/examples/train/sequence_parallel)。
- 🎁 2025.05.11: GRPO中的奖励模型支持自定义处理逻辑，GenRM的例子参考[这里](./docs/source/Instruction/GRPO/DeveloperGuide/reward_model.md)。
Expand Down
Loading
Loading