[PR #13063] Add deepseek v3.1 #12793

Open
opened 2025-11-12 17:06:51 -06:00 by GiteaMirror · 0 comments
Owner

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/13063
Author: @gr4ceG
Created: 11/12/2025
Status: 🔄 Open

Base: mainHead: grace/deepseek-v3.1-update


📝 Commits (7)

  • 09bdac6 splitting kvb
  • 8c298b8 Changed ggml backend to support mla
  • 1febcb4 working with deepseekv3.1 thinking - sdpa + non-flash
  • 8336f51 works for 3.1, but regression in 3???
  • dd76959 working v3 and v3.1 with conditional branches
  • 46519bc cleanup
  • 99cc655 cleanup

📊 Changes

4 files changed (+66 additions, -32 deletions)

View changed files

📝 ml/backend.go (+2 -1)
📝 ml/backend/ggml/ggml.go (+8 -4)
📝 ml/nn/attention.go (+11 -2)
📝 model/models/deepseek2/model.go (+45 -25)

📄 Description

TODO: working with split tensors


🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.

## 📋 Pull Request Information **Original PR:** https://github.com/ollama/ollama/pull/13063 **Author:** [@gr4ceG](https://github.com/gr4ceG) **Created:** 11/12/2025 **Status:** 🔄 Open **Base:** `main` ← **Head:** `grace/deepseek-v3.1-update` --- ### 📝 Commits (7) - [`09bdac6`](https://github.com/ollama/ollama/commit/09bdac633201b07b9ef455a518919608e99b3582) splitting kvb - [`8c298b8`](https://github.com/ollama/ollama/commit/8c298b836813c289c53fbdc683cc05d9119f2d91) Changed ggml backend to support mla - [`1febcb4`](https://github.com/ollama/ollama/commit/1febcb40285377d2cd0ae0c6b8a06e0f1f62efc7) working with deepseekv3.1 thinking - sdpa + non-flash - [`8336f51`](https://github.com/ollama/ollama/commit/8336f5195576d1428730846018dde73a1d59a834) works for 3.1, but regression in 3??? - [`dd76959`](https://github.com/ollama/ollama/commit/dd76959a1c2ab85e13d852ecfb5982791f9e0f58) working v3 and v3.1 with conditional branches - [`46519bc`](https://github.com/ollama/ollama/commit/46519bc7f581340a816988545eefa6f1724ff67d) cleanup - [`99cc655`](https://github.com/ollama/ollama/commit/99cc65549e0e2a9c9dd8ac5c06f05b54a8cbc8a6) cleanup ### 📊 Changes **4 files changed** (+66 additions, -32 deletions) <details> <summary>View changed files</summary> 📝 `ml/backend.go` (+2 -1) 📝 `ml/backend/ggml/ggml.go` (+8 -4) 📝 `ml/nn/attention.go` (+11 -2) 📝 `model/models/deepseek2/model.go` (+45 -25) </details> ### 📄 Description TODO: working with split tensors --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
GiteaMirror added the
pull-request
label 2025-11-12 17:06:51 -06:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama-ollama#12793
No description provided.