github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[PR #7767] [MERGED] KV Cache Fixes #17783

New Issue
Closed
opened 2026-04-16 06:13:52 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-16 06:13:52 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/7767
Author: @jessegross
Created: 11/20/2024
Status: ✅ Merged
Merged: 11/20/2024
Merged by: @jessegross

Base: main ← Head: jessegross/kv

📝 Commits (5)

  • e1a72ba runner.go: Use correct index when retrieving embedding results
  • f34bbe2 runner.go: Retry decoding after defragmentation if needed
  • 065451f runner.go: Hard fail on errors rather than potentially infinite looping
  • 85a9b21 runner.go: Don't add inputs to cache view until actually processed
  • 26e0461 runner.go: Truncate inputs that exceed context rather than shifting

📊 Changes

4 files changed (+94 additions, -37 deletions)

View changed files

📝 integration/context_test.go (+31 -0)
📝 llama/llama.go (+10 -4)
📝 llama/runner/cache.go (+12 -4)
📝 llama/runner/runner.go (+41 -29)

📄 Description

Users have reported a number of errors related to the KV cache such as:

  • Error: "could not find a KV slot for the batch - try reducing the size of the batch or increase the context. code: 1"
  • Hanging due to infinite loops
  • Output that ends unexpectedly
  • Slower performance than before when passing inputs that are much longer than the context size

This aims to both fix these problems and continue to make this area of the code less error prone.

🔄 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/7767 **Author:** [@jessegross](https://github.com/jessegross) **Created:** 11/20/2024 **Status:** ✅ Merged **Merged:** 11/20/2024 **Merged by:** [@jessegross](https://github.com/jessegross) **Base:** `main` ← **Head:** `jessegross/kv` --- ### 📝 Commits (5) - [`e1a72ba`](https://github.com/ollama/ollama/commit/e1a72bacb49120fd8cdbe07e6ae24f4a6c58cf1d) runner.go: Use correct index when retrieving embedding results - [`f34bbe2`](https://github.com/ollama/ollama/commit/f34bbe2d0d26a405f0d4135fa3cbb17654b16ecd) runner.go: Retry decoding after defragmentation if needed - [`065451f`](https://github.com/ollama/ollama/commit/065451f66889d8af9d01f8b768c73ef4d5f9cf7f) runner.go: Hard fail on errors rather than potentially infinite looping - [`85a9b21`](https://github.com/ollama/ollama/commit/85a9b2126c30dc429ea17a7971b3f3ed085b697f) runner.go: Don't add inputs to cache view until actually processed - [`26e0461`](https://github.com/ollama/ollama/commit/26e0461d132f4ab6e5d493f971c08ce9299b5758) runner.go: Truncate inputs that exceed context rather than shifting ### 📊 Changes **4 files changed** (+94 additions, -37 deletions) <details> <summary>View changed files</summary> 📝 `integration/context_test.go` (+31 -0) 📝 `llama/llama.go` (+10 -4) 📝 `llama/runner/cache.go` (+12 -4) 📝 `llama/runner/runner.go` (+41 -29) </details> ### 📄 Description Users have reported a number of errors related to the KV cache such as: - Error: "could not find a KV slot for the batch - try reducing the size of the batch or increase the context. code: 1" - Hanging due to infinite loops - Output that ends unexpectedly - Slower performance than before when passing inputs that are much longer than the context size This aims to both fix these problems and continue to make this area of the code less error prone. --- <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 2026-04-16 06:13:52 -05:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label pull-request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#17783
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?