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 #8392] Empty 'assistant' message #31148

New Issue
Closed
opened 2026-04-22 11:20:15 -05:00 by GiteaMirror · 4 comments
GiteaMirror commented 2026-04-22 11:20:15 -05:00
Owner
Copy Link

Originally created by @pulinagrawal on GitHub (Jan 12, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/8392

What is the issue?

with the following python code

>>> import ollama
>>> my = {
...   "model": "llama3.2",
...   "options": {
...     "temperature": 0
...   },
...   "messages": [{"role": "system", "content": "You are a DnD Dungeon Master. Say something in your first message to the user."}
...   ],
...   "tools": [
...       {
...         "type": "function",
...         "function": {
...           "name": "roll_for_action",
...           "description": "Checks a dice roll for a skill with a given difficulty class",
...           "parameters": {
...               "type": "object",
...               "properties": {"n_dice": {"type": "integer"},
...                              "sides": {"type": "integer"},
...                              "skill": {"type": "string"},
...                              "dc": {"type": "integer"},
...                              "player": {"type": "string"}
...                             }
...           }
...         }
...       }]
... }
>>> response = ollama.chat(**my)
>>> print(response)
model='llama3.2' created_at='2025-01-12T06:38:51.757532Z' done=True done_reason='stop' total_duration=1349066625 load_duration=16419958 prompt_eval_count=66 prompt_eval_duration=1330000000 eval_count=1 eval_duration=1000000 message=Message(role='assistant', content='', images=None, tool_calls=None)

The response content is empty.

OS

macOS

GPU

Apple

CPU

Apple

Ollama version

0.5.4

Originally created by @pulinagrawal on GitHub (Jan 12, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/8392 ### What is the issue? with the following python code ``` >>> import ollama >>> my = { ... "model": "llama3.2", ... "options": { ... "temperature": 0 ... }, ... "messages": [{"role": "system", "content": "You are a DnD Dungeon Master. Say something in your first message to the user."} ... ], ... "tools": [ ... { ... "type": "function", ... "function": { ... "name": "roll_for_action", ... "description": "Checks a dice roll for a skill with a given difficulty class", ... "parameters": { ... "type": "object", ... "properties": {"n_dice": {"type": "integer"}, ... "sides": {"type": "integer"}, ... "skill": {"type": "string"}, ... "dc": {"type": "integer"}, ... "player": {"type": "string"} ... } ... } ... } ... }] ... } >>> response = ollama.chat(**my) >>> print(response) model='llama3.2' created_at='2025-01-12T06:38:51.757532Z' done=True done_reason='stop' total_duration=1349066625 load_duration=16419958 prompt_eval_count=66 prompt_eval_duration=1330000000 eval_count=1 eval_duration=1000000 message=Message(role='assistant', content='', images=None, tool_calls=None) ``` The response content is empty. ### OS macOS ### GPU Apple ### CPU Apple ### Ollama version 0.5.4
GiteaMirror added the bug label 2026-04-22 11:20:15 -05:00
GiteaMirror commented 2026-04-22 11:20:18 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jan 12, 2025):

The system message is generally used for setting the tone, supplying tools, etc and not direct instruction. LLMs are trained to perform generations from a user or assistant message, since none is supplied, the LLM has no response.

<!-- gh-comment-id:2585618786 --> @rick-github commented on GitHub (Jan 12, 2025): The system message is generally used for setting the tone, supplying tools, etc and not direct instruction. LLMs are trained to perform generations from a user or assistant message, since none is supplied, the LLM has no response.
GiteaMirror commented 2026-04-22 11:20:18 -05:00
Author
Owner
Copy Link

@pulinagrawal commented on GitHub (Jan 12, 2025):

That doesn't make sense. They are not trained to provide a welcome greeting message? I will be surprised if that is how they are created. I checked for a similar message with only a system prompt to ChatGPT. It produces an assistant message.

Anyway thanks for your response @rick-github.

By the way I was able to fix this issue by changing the template in Modelfile like so:

{{- end }}{{ if $last }}<|start_header_id|>assistant<|end_header_id|>

{{ end }}
{{- else if eq .Role "system" }} <|start_header_id|>assistant<|end_header_id|>

{{- else if eq .Role "assistant" }}<|start_header_id|>assistant<|end_header_id|>

I added {{- else if eq .Role "system" }} <|start_header_id|>assistant<|end_header_id|> on line 35.

<!-- gh-comment-id:2585636489 --> @pulinagrawal commented on GitHub (Jan 12, 2025): That doesn't make sense. They are not trained to provide a welcome greeting message? I will be surprised if that is how they are created. I checked for a similar message with only a system prompt to ChatGPT. It produces an assistant message. Anyway thanks for your response @rick-github. By the way I was able to fix this issue by changing the template in Modelfile like so: ``` {{- end }}{{ if $last }}<|start_header_id|>assistant<|end_header_id|> {{ end }} {{- else if eq .Role "system" }} <|start_header_id|>assistant<|end_header_id|> {{- else if eq .Role "assistant" }}<|start_header_id|>assistant<|end_header_id|> ``` I added ```{{- else if eq .Role "system" }} <|start_header_id|>assistant<|end_header_id|>``` on line 35.
GiteaMirror commented 2026-04-22 11:20:19 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jan 12, 2025):

Modifying the template is not necessary, adding a blank assistant message to your prompt is enough.

<!-- gh-comment-id:2585645388 --> @rick-github commented on GitHub (Jan 12, 2025): Modifying the template is not necessary, adding a blank assistant message to your prompt is enough.
GiteaMirror commented 2026-04-22 11:20:22 -05:00
Author
Owner
Copy Link

@pulinagrawal commented on GitHub (Jan 12, 2025):

Thanks @rick-github that is a better solution.

close #8392

<!-- gh-comment-id:2585795163 --> @pulinagrawal commented on GitHub (Jan 12, 2025): Thanks @rick-github that is a better solution. close #8392
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 bug
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#31148
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?