github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 16:11:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[PR #9973] server: support streaming near tool usage #59807

New Issue
Open
opened 2026-04-29 14:44:33 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-29 14:44:33 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/9973
Author: @fizx
Created: 3/25/2025
Status: 🔄 Open

Base: main ← Head: streaming

📝 Commits (3)

📊 Changes

3 files changed (+103 additions, -21 deletions)

View changed files

📝 server/images.go (+51 -0)
📝 server/model_test.go (+34 -16)
📝 server/routes.go (+18 -5)

📄 Description

Right now, including any tools in the ChatRequest implicitly turns off streaming responses.

While we don't want to stream tool calls, we often want to stream the text
adjacent to those tool calls. This patch looks at the response and selectively engages streaming.

First PR, trying to follow guidelines, but YMMV. Feedback welcome.

curl localhost:11434/api/chat -d @test.json

Before:

% curl localhost:11434/api/chat -d @test.json
{"model":"llama3.3:latest","created_at":"2025-03-25T05:51:16.13806Z","message":{"role":"assistant","content":"The answer to 2+2 is 4."},"done_reason":"stop","done":true,"total_duration":2643049709,"load_duration":15155625,"prompt_eval_count":92,"prompt_eval_duration":1516872875,"eval_count":12,"eval_duration":1110346291}

After:

% curl localhost:11434/api/chat -d @test.json
{"model":"llama3.3:latest","created_at":"2025-03-25T05:54:36.602815Z","message":{"role":"assistant","content":"The"},"done":false}
{"model":"llama3.3:latest","created_at":"2025-03-25T05:54:36.702575Z","message":{"role":"assistant","content":" answer"},"done":false}
{"model":"llama3.3:latest","created_at":"2025-03-25T05:54:36.80337Z","message":{"role":"assistant","content":" to"},"done":false}
{"model":"llama3.3:latest","created_at":"2025-03-25T05:54:36.904238Z","message":{"role":"assistant","content":" "},"done":false}
{"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.005213Z","message":{"role":"assistant","content":"2"},"done":false}
{"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.106363Z","message":{"role":"assistant","content":"+"},"done":false}
{"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.207305Z","message":{"role":"assistant","content":"2"},"done":false}
{"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.308348Z","message":{"role":"assistant","content":" is"},"done":false}
{"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.409389Z","message":{"role":"assistant","content":" "},"done":false}
{"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.5104Z","message":{"role":"assistant","content":"4"},"done":false}
{"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.611554Z","message":{"role":"assistant","content":"."},"done":false}
{"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.712799Z","message":{"role":"assistant","content":""},"done_reason":"stop","done":true,"total_duration":3462526792,"load_duration":1156274000,"prompt_eval_count":92,"prompt_eval_duration":1194830875,"eval_count":12,"eval_duration":1110376750}

🔄 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/9973 **Author:** [@fizx](https://github.com/fizx) **Created:** 3/25/2025 **Status:** 🔄 Open **Base:** `main` ← **Head:** `streaming` --- ### 📝 Commits (3) - [`b4ae2de`](https://github.com/ollama/ollama/commit/b4ae2de45a23a0786bd172f3bfe6c671188a2aa2) add a prefix to the model - [`c16c813`](https://github.com/ollama/ollama/commit/c16c813736812b15f1b145255d3dd89660324a91) finish a streaming patch - [`e5512a6`](https://github.com/ollama/ollama/commit/e5512a6e658233ae468ab4919953262d0a6b792c) appease the linter ### 📊 Changes **3 files changed** (+103 additions, -21 deletions) <details> <summary>View changed files</summary> 📝 `server/images.go` (+51 -0) 📝 `server/model_test.go` (+34 -16) 📝 `server/routes.go` (+18 -5) </details> ### 📄 Description Right now, including any tools in the ChatRequest implicitly turns off streaming responses. While we don't want to stream tool calls, we often want to stream the text adjacent to those tool calls. This patch looks at the response and selectively engages streaming. First PR, trying to follow guidelines, but YMMV. Feedback welcome. curl localhost:11434/api/chat -d @[test.json](https://github.com/user-attachments/files/19442584/test.json) Before: ``` % curl localhost:11434/api/chat -d @test.json {"model":"llama3.3:latest","created_at":"2025-03-25T05:51:16.13806Z","message":{"role":"assistant","content":"The answer to 2+2 is 4."},"done_reason":"stop","done":true,"total_duration":2643049709,"load_duration":15155625,"prompt_eval_count":92,"prompt_eval_duration":1516872875,"eval_count":12,"eval_duration":1110346291} ``` After: ``` % curl localhost:11434/api/chat -d @test.json {"model":"llama3.3:latest","created_at":"2025-03-25T05:54:36.602815Z","message":{"role":"assistant","content":"The"},"done":false} {"model":"llama3.3:latest","created_at":"2025-03-25T05:54:36.702575Z","message":{"role":"assistant","content":" answer"},"done":false} {"model":"llama3.3:latest","created_at":"2025-03-25T05:54:36.80337Z","message":{"role":"assistant","content":" to"},"done":false} {"model":"llama3.3:latest","created_at":"2025-03-25T05:54:36.904238Z","message":{"role":"assistant","content":" "},"done":false} {"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.005213Z","message":{"role":"assistant","content":"2"},"done":false} {"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.106363Z","message":{"role":"assistant","content":"+"},"done":false} {"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.207305Z","message":{"role":"assistant","content":"2"},"done":false} {"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.308348Z","message":{"role":"assistant","content":" is"},"done":false} {"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.409389Z","message":{"role":"assistant","content":" "},"done":false} {"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.5104Z","message":{"role":"assistant","content":"4"},"done":false} {"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.611554Z","message":{"role":"assistant","content":"."},"done":false} {"model":"llama3.3:latest","created_at":"2025-03-25T05:54:37.712799Z","message":{"role":"assistant","content":""},"done_reason":"stop","done":true,"total_duration":3462526792,"load_duration":1156274000,"prompt_eval_count":92,"prompt_eval_duration":1194830875,"eval_count":12,"eval_duration":1110376750} ``` --- <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 2026-04-29 14:44:33 -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#59807
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?