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

[GH-ISSUE #6689] Reflection 70B fix? #66246

New Issue
Closed
opened 2026-05-04 01:22:39 -05:00 by GiteaMirror · 5 comments
GiteaMirror commented 2026-05-04 01:22:39 -05:00
Owner
Copy Link

Originally created by @gileneusz on GitHub (Sep 7, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/6689

There are rumors that Reflection model does not run properly on ollama, can anyone confirm it?

comments here:

https://www.reddit.com/r/LocalLLaMA/comments/1fa72an/reflectionllama3170b_available_on_ollama/

Originally created by @gileneusz on GitHub (Sep 7, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/6689 There are rumors that Reflection model does not run properly on ollama, can anyone confirm it? comments here: https://www.reddit.com/r/LocalLLaMA/comments/1fa72an/reflectionllama3170b_available_on_ollama/
GiteaMirror added the model label 2026-05-04 01:22:40 -05:00
GiteaMirror commented 2026-05-04 01:22:56 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Sep 7, 2024):

I haven't used it much because I don't have the hardware to run at a reasonable speed, but it seems to have basic functionality.

$ ollama run reflection:70b-q4_0
>>> hello
<thinking>
To respond to "hello", I'll need to consider several factors:

1. The user's intent: They're likely saying hello as a greeting.
2. Appropriate responses: Common ways to respond to "hello" are:
   - Hello back
   - Hi
   - Hey
3. Formality level: Since this is an AI response, I'll aim for a friendly but professional tone.

<reflection>
The approach of responding with a standard greeting seems appropriate in this context. It acknowledges the user's hello and provides a polite response.
</reflection>

Given these considerations, the most suitable response would be to echo "hello" back to the user.
</thinking>

<output>
Hello!
</output>

>>> Send a message (/? for help)
<!-- gh-comment-id:2336398183 --> @rick-github commented on GitHub (Sep 7, 2024): I haven't used it much because I don't have the hardware to run at a reasonable speed, but it seems to have basic functionality. ``` $ ollama run reflection:70b-q4_0 >>> hello <thinking> To respond to "hello", I'll need to consider several factors: 1. The user's intent: They're likely saying hello as a greeting. 2. Appropriate responses: Common ways to respond to "hello" are: - Hello back - Hi - Hey 3. Formality level: Since this is an AI response, I'll aim for a friendly but professional tone. <reflection> The approach of responding with a standard greeting seems appropriate in this context. It acknowledges the user's hello and provides a polite response. </reflection> Given these considerations, the most suitable response would be to echo "hello" back to the user. </thinking> <output> Hello! </output> >>> Send a message (/? for help) ```
GiteaMirror commented 2026-05-04 01:23:01 -05:00
Author
Owner
Copy Link

@gileneusz commented on GitHub (Sep 7, 2024):

I had the same - good outputs, but benchmarks results were inferior to llama 3.1 70b...

<!-- gh-comment-id:2336398516 --> @gileneusz commented on GitHub (Sep 7, 2024): I had the same - good outputs, but benchmarks results were inferior to llama 3.1 70b...
GiteaMirror commented 2026-05-04 01:23:06 -05:00
Author
Owner
Copy Link

@gileneusz commented on GitHub (Sep 8, 2024):

new weights just dropped:

https://huggingface.co/mattshumer/ref_70_e3/tree/main

<!-- gh-comment-id:2336817052 --> @gileneusz commented on GitHub (Sep 8, 2024): new weights just dropped: https://huggingface.co/mattshumer/ref_70_e3/tree/main
GiteaMirror commented 2026-05-04 01:23:12 -05:00
Author
Owner
Copy Link

@DougHaber commented on GitHub (Sep 8, 2024):

I was wondering the same thing at first. My attempts didn't go well. To save anyone else a few minutes, this model doesn't work correctly without the system prompt. If you are overriding that or using it through the API, it might be necessary to manually provide it. On X, Matt Shumer said removing that requirement is on his list for a future version. Even with that, the original released weights were broken. I haven't had a chance to try the updated weights yet.

<!-- gh-comment-id:2336859616 --> @DougHaber commented on GitHub (Sep 8, 2024): I was wondering the same thing at first. My attempts didn't go well. To save anyone else a few minutes, this model doesn't work correctly without the [system prompt](https://ollama.com/library/reflection/blobs/c782799f4377). If you are overriding that or using it through the API, it might be necessary to manually provide it. On X, [Matt Shumer said](https://x.com/mattshumer_/status/1832169489309561309) removing that requirement is on his list for a future version. Even with that, the original released weights were broken. I haven't had a chance to try the updated weights yet.
GiteaMirror commented 2026-05-04 01:23:18 -05:00
Author
Owner
Copy Link

@gileneusz commented on GitHub (Sep 8, 2024):

they benchmarked this on claude, and published fine tuned llama 3.... Matt missed it somehow...

https://www.reddit.com/r/LocalLLaMA/comments/1fc98fu/confirmed_reflection_70bs_official_api_is_sonnet/

<!-- gh-comment-id:2336862080 --> @gileneusz commented on GitHub (Sep 8, 2024): they benchmarked this on claude, and published fine tuned llama 3.... Matt missed it somehow... https://www.reddit.com/r/LocalLLaMA/comments/1fc98fu/confirmed_reflection_70bs_official_api_is_sonnet/
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 model
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#66246
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?