Error rendering prompt with jinja template: TypeError: Cannot read properties of undefined (reading 'type')

[fasthold]

LM Studio: 0.3.8
Platform: MacBook Pro M3 Pro
OS: macOS 15.1.1
Model: qwen2.5-coder-32b-instruct & deepseek-r1-distill-qwen-7b

Got the error below:

2025-01-26 19:45:53  [INFO] 
[LM STUDIO SERVER] Running chat completion on conversation with 2 messages.
2025-01-26 19:45:53  [INFO] 
[LM STUDIO SERVER] Streaming response...
2025-01-26 19:45:53 [DEBUG] 
[createJinjaMessagesInput] Ignoring image file part in user message due to missing image config
2025-01-26 19:45:53 [DEBUG] 
1 Error predicting: Error: Error rendering prompt with jinja template: TypeError: Cannot read properties of undefined (reading 'type')
    at /Applications/LM Studio.app/Contents/Resources/app/.webpack/lib/llmworker.js:9:61010
    at async _0x19f928.<computed> (/Applications/LM Studio.app/Contents/Resources/app/.webpack/lib/llmworker.js:9:56076)
    at async _0x47a04d.<computed> (/Applications/LM Studio.app/Contents/Resources/app/.webpack/lib/llmworker.js:9:81857)
    at async _0x2800dc.temp_createLlamaPredictionArgs (/Applications/LM Studio.app/Contents/Resources/app/.webpack/lib/llmworker.js:7:761918)
    at async _0x3ee58d.<computed>.predictTokens (/Applications/LM Studio.app/Contents/Resources/app/.webpack/lib/llmworker.js:9:43800)
    at async Object.predictTokens (/Applications/LM Studio.app/Contents/Resources/app/.webpack/lib/llmworker.js:14:10469)
    at async Object.handleMessage (/Applications/LM Studio.app/Contents/Resources/app/.webpack/lib/llmworker.js:14:1973)
2025-01-26 19:45:53 [ERROR] 
Error rendering prompt with jinja template: TypeError: Cannot read properties of undefined (reading 'type'). Error Data: n/a, Additional Data: n/a

[sumeet-kg]

SIMILAR issue with linux

LM Studio: 0.3.8
platform: Linux with Ubuntu 24.04 LTS
Model: DeepSeek-R1-Distill-Qwen-7B-GGUF/DeepSeek-R1-Distill-Qwen-7B-Q4_K_M.gguf

Please help
Error code: 400 - {'error': "Error rendering prompt with jinja template: TypeError: Cannot read properties of undefined (reading 'type')"}

[MorpheusMalatorus]

Similar issue with Windows 10

LM Studio 0.3.9
Model: mradermacher/Mistral-Small-24B-Instruct-2501.i1-IQ3_M.gguf

"Error rendering prompt with jinja template: Error: Cannot call something that is not a function: got UndefinedValue"

[yagil]

There are actually 2 different bugs reported here.

1. TypeError: Cannot read properties of undefined (reading 'type') -> this is fixed in 0.3.9 build 6
2. Error: Cannot call something that is not a function: got UndefinedValue -> this is due to LM Studio not supporting strftime in jinja templates, which we should fix. You can get around it in the meantime by editing the jinja and removing that part

[dtm-beep]

I wasn't able to find the offending part relating to using Mistral-Small-24B-Instruct-2501 for

Error: Cannot call something that is not a function: got UndefinedValue -> this is due to LM Studio not supporting strftime in jinja templates, which we should fix. You can get around it in the meantime by editing the jinja and removing that part

But I was able to run the model by selecting the Llama 3 preset instead in My Models / Prompt template / Manual -> Llama 3.

[phazei]

Same problem: mistral-small-24b-instruct-2501-abliterated

@dtm-beep the Llama 3 trick worked

[nekogyo]

run the model by selecting the Llama 3 preset instead in My Models / Prompt template / Manual -> Llama 3.

尝试可行，在我的模型把提示词模板从jinja改为Llama 3，可正常使用。

[mptm436]

There are actually 2 different bugs reported here.

1. TypeError: Cannot read properties of undefined (reading 'type') -> this is fixed in 0.3.9 build 6
2. Error: Cannot call something that is not a function: got UndefinedValue -> this is due to LM Studio not supporting strftime in jinja templates, which we should fix. You can get around it in the meantime by editing the jinja and removing that part

Same problem: mistral-small-24b-instruct-2501-abliterated

@dtm-beep the Llama 3 trick worked

For mistral model, delete {%- set today = strftime_now("%Y-%m-%d") %} and The current date is " + today + ".\n\n two sentences from the jinja prompt would work

[itsfarseen]

I fixed this for granite 2 by adding a system prompt in the system prompt text box.

[42531ljm]

run the model by selecting the Llama 3 preset instead in My Models / Prompt template / Manual -> Llama 3.

尝试可行，在我的模型把提示词模板从jinja改为Llama 3，可正常使用。

works for qwen_qwq

[serhii-nakon]

Hello I also trying to fix this issue with llama-cpp-python - and here fix abetlen/llama-cpp-python#1486