I have searched for any existing and/or related issues.
I have searched for any existing and/or related discussions.
I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!).
I am using the latest version of Open WebUI.
Installation Method
Pip Install
Open WebUI Version
0.8.5
Ollama Version (if applicable)
No response
Operating System
MacOS 26.3.1
Browser (if applicable)
Safari
Confirmation
I have read and followed all instructions in README.md.
I am using the latest version of both Open WebUI and Ollama.
I have included the browser console logs.
I have included the Docker container logs.
I have provided every relevant configuration, setting, and environment variable used in my setup.
I have clearly listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc).
I have documented step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation. My steps:
Start with the initial platform/version/OS and dependencies used,
Specify exact install/launch/configure commands,
List URLs visited, user input (incl. example values/emails/passwords if needed),
Describe all options and toggles enabled or changed,
Include any files or environmental changes,
Identify the expected and actual result at each stage,
Ensure any reasonably skilled user can follow and hit the same issue.
Expected Behavior
pasting 5432:100, 413:-100 should make model prefer some tokens more, tried different ways to set it including the exact same what mentioned in placeholder:
placeholder="Enter comma-separated "token:bias_value" pairs (example: 5432:100, 413:-100)"
Actual Behavior
'dict' object has no attribute 'split'
or some different errors, also tried patching convert_logit_bias_input_to_json function and this one worked for me:
def convert_logit_bias_input_to_json(user_input) -> Optional[str]:
if not user_input:
return None
# FIX: already parsed as dict
if isinstance(user_input, dict):
return json.dumps(user_input)
logit_bias_pairs = user_input.split(",")
logit_bias_json = {}
for pair in logit_bias_pairs:
token, bias = pair.split(":")
token = str(token.strip())
bias = int(bias.strip())
bias = 100 if bias > 100 else -100 if bias < -100 else bias
logit_bias_json[token] = bias
return json.dumps(logit_bias_json)
Steps to Reproduce
set parameters to something like this and use backend that supports logit_bias like llama.cpp
Logs & Screenshots
Additional Information
No response
Originally created by @TTTap1xxx on GitHub (Mar 11, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/22597
### Check Existing Issues
- [x] I have searched for any existing and/or related issues.
- [x] I have searched for any existing and/or related discussions.
- [x] I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!).
- [x] I am using the latest version of Open WebUI.
### Installation Method
Pip Install
### Open WebUI Version
0.8.5
### Ollama Version (if applicable)
_No response_
### Operating System
MacOS 26.3.1
### Browser (if applicable)
Safari
### Confirmation
- [x] I have read and followed all instructions in `README.md`.
- [x] I am using the latest version of **both** Open WebUI and Ollama.
- [x] I have included the browser console logs.
- [x] I have included the Docker container logs.
- [x] I have **provided every relevant configuration, setting, and environment variable used in my setup.**
- [x] I have clearly **listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup** (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc).
- [x] I have documented **step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation**. My steps:
- Start with the initial platform/version/OS and dependencies used,
- Specify exact install/launch/configure commands,
- List URLs visited, user input (incl. example values/emails/passwords if needed),
- Describe all options and toggles enabled or changed,
- Include any files or environmental changes,
- Identify the expected and actual result at each stage,
- Ensure any reasonably skilled user can follow and hit the same issue.
### Expected Behavior
pasting 5432:100, 413:-100 should make model prefer some tokens more, tried different ways to set it including the exact same what mentioned in placeholder:
placeholder="Enter comma-separated "token:bias_value" pairs (example: 5432:100, 413:-100)"
### Actual Behavior
'dict' object has no attribute 'split'
or some different errors, also tried patching convert_logit_bias_input_to_json function and this one worked for me:
def convert_logit_bias_input_to_json(user_input) -> Optional[str]:
if not user_input:
return None
# FIX: already parsed as dict
if isinstance(user_input, dict):
return json.dumps(user_input)
logit_bias_pairs = user_input.split(",")
logit_bias_json = {}
for pair in logit_bias_pairs:
token, bias = pair.split(":")
token = str(token.strip())
bias = int(bias.strip())
bias = 100 if bias > 100 else -100 if bias < -100 else bias
logit_bias_json[token] = bias
return json.dumps(logit_bias_json)
### Steps to Reproduce
<img width="1386" height="275" alt="Image" src="https://github.com/user-attachments/assets/e6b08247-3a81-4ffe-b60f-390bba7e5791" />
set parameters to something like this and use backend that supports logit_bias like llama.cpp
<img width="1386" height="275" alt="Image" src="https://github.com/user-attachments/assets/12b19f68-9907-4a0c-b1dd-544a328b2d9e" />
### Logs & Screenshots
<img width="1386" height="275" alt="Image" src="https://github.com/user-attachments/assets/70f47e1b-3d70-4be2-abb3-5c221f6d0ecb" />
<img width="1386" height="275" alt="Image" src="https://github.com/user-attachments/assets/8335a0b7-8774-4a70-b754-bfad654f79da" />
### Additional Information
_No response_
GiteaMirror
added the bug label 2026-04-25 09:31:21 -05:00
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Originally created by @TTTap1xxx on GitHub (Mar 11, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/22597
Check Existing Issues
Installation Method
Pip Install
Open WebUI Version
0.8.5
Ollama Version (if applicable)
No response
Operating System
MacOS 26.3.1
Browser (if applicable)
Safari
Confirmation
README.md.Expected Behavior
pasting 5432:100, 413:-100 should make model prefer some tokens more, tried different ways to set it including the exact same what mentioned in placeholder:
placeholder="Enter comma-separated "token:bias_value" pairs (example: 5432:100, 413:-100)"
Actual Behavior
'dict' object has no attribute 'split'
or some different errors, also tried patching convert_logit_bias_input_to_json function and this one worked for me:
def convert_logit_bias_input_to_json(user_input) -> Optional[str]:
if not user_input:
return None
Steps to Reproduce
Logs & Screenshots
Additional Information
No response
@tjbck commented on GitHub (Mar 11, 2026):
e34ed72e1e