Open susravya-d opened 1 month ago
It looks like the issue might be related to your plain_text_resume.yaml
configuration. Please check the legal_authorization
section in your yaml file and make sure it's set correctly.
legal_authorization:
eu_work_authorization: "No"
us_work_authorization: "Yes"
requires_us_visa: "No"
requires_us_sponsorship: "No"
requires_eu_visa: "Yes"
legally_allowed_to_work_in_eu: "No"
legally_allowed_to_work_in_us: "Yes"
requires_eu_sponsorship: "Yes"
canada_work_authorization: "No"
requires_canada_visa: "Yes"
legally_allowed_to_work_in_canada: "No"
requires_canada_sponsorship: "Yes"
uk_work_authorization: "No"
requires_uk_visa: "Yes"
legally_allowed_to_work_in_uk: "No"
requires_uk_sponsorship: "Yes"
This is my configuration.
I also hit this bug. I have turned the temperature from 0.4 to 0.7 and and answered all the Canada, UK, and US work questions the same, e.g. no visa, yes auth, etc.
assumption that gemini LLM is hallucinating on the information, so the above modification.
I also hit this bug. I have turned the temperature from 0.4 to 0.7 and and answered all the Canada, UK, and US work questions the same, e.g. no visa, yes auth, etc.
assumption that gemini LLM is hallucinating on the information, so the above modification.
I don't think this is the problem. I've tried Gemini, GPT4o, and (Ollama) Llama 3.2. These all seem to result in the same inaccuracies. I believe that the bot is ignoring parameters in general. It's like the code that tells the LLM each field and what they are is missing something. Also, when editing the configs it seems to be ignoring certain changes, or perhaps it was never reading the information anyways.
Go into answers.json and fix up there. These do get reviewed as part of the questions answer process. Sean @. On Oct 22, 2024, at 08:45, sonicnerd14 @.> wrote:
I also hit this bug. I have turned the temperature from 0.4 to 0.7 and and answered all the Canada, UK, and US work questions the same, e.g. no visa, yes auth, etc. assumption that gemini LLM is hallucinating on the information, so the above modification.
I don't think this is the problem. I've tried Gemini, GPT4o, and (Ollama) Llama 3.2. These all seem to result in the same inaccuracies. I believe that the bot is ignoring parameters in general. It's like the code that tells the LLM each field and what they are is missing something. Also, when editing the configs it seems to be ignoring certain changes, or perhaps it was never reading the information anyways.
—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you commented.Message ID: @.***>
Issue description
I don't know how it is choosing the answers. I want to answer yes to this question but it is submitting no. Here is the debug logs.
2024-10-14 00:46:24.786 | DEBUG | src.aihawk_easy_applier:_fill_additional_questions:605 - Filling additional questions 2024-10-14 00:46:24.790 | DEBUG | src.aihawk_easy_applier:_process_form_section:611 - Processing form section 2024-10-14 00:46:24.825 | DEBUG | src.aihawk_easy_applier:_sanitize_text:849 - Sanitized text: are you legally authorized to work in the united states?are you legally authorized to work in the united states?requiredyesno 2024-10-14 00:46:24.825 | DEBUG | src.aihawk_easy_applier:_sanitize_text:849 - Sanitized text: are you legally authorized to work in the united states?are you legally authorized to work in the united states?requiredyesno 2024-10-14 00:46:24.825 | DEBUG | src.aihawk_easy_applier:_sanitize_text:849 - Sanitized text: are you legally authorized to work in the united states?are you legally authorized to work in the united states?requiredyesno 2024-10-14 00:46:24.825 | DEBUG | src.aihawk_easy_applier:_sanitize_text:849 - Sanitized text: are you legally authorized to work in the united states?are you legally authorized to work in the united states?requiredyesno 2024-10-14 00:46:24.825 | DEBUG | src.aihawk_easy_applier:_sanitize_text:849 - Sanitized text: are you legally authorized to work in the united states?are you legally authorized to work in the united states?requiredyesno 2024-10-14 00:46:24.825 | DEBUG | src.aihawk_easy_applier:_sanitize_text:849 - Sanitized text: are you legally authorized to work in the united states?are you legally authorized to work in the united states?requiredyesno 2024-10-14 00:46:24.825 | DEBUG | src.aihawk_easy_applier:_sanitize_text:849 - Sanitized text: are you legally authorized to work in the united states?are you legally authorized to work in the united states?requiredyesno 2024-10-14 00:46:24.825 | DEBUG | src.aihawk_easy_applier:_sanitize_text:849 - Sanitized text: are you legally authorized to work in the united states?are you legally authorized to work in the united states?requiredyesno 2024-10-14 00:46:24.825 | DEBUG | src.aihawk_easy_applier:_sanitize_text:849 - Sanitized text: are you legally authorized to work in the united states?are you legally authorized to work in the united states?requiredyesno 2024-10-14 00:46:24.825 | DEBUG | src.aihawk_easy_applier:_select_radio:808 - Selecting radio option: no 2024-10-14 00:46:24.914 | DEBUG | src.aihawk_easy_applier:_find_and_handle_radio_question:653 - Selected existing radio answer 2024-10-14 00:46:24.925 | DEBUG | src.aihawk_easy_applier:_process_form_section:616 - Handled radio question 2024-10-14 00:46:24.925 | DEBUG | src.aihawk_easy_applier:_process_form_section:611 - Processing form section
How to fix this?
Specific tasks
No response
Additional resources
No response
Additional context
No response