You can deploy this model and start earning money today!

michaelarutyunov/
jtbd-t2l-jobtrigger

safetensors

Views

No views yet

T2L Adapter: job_trigger

Text-to-LoRA (T2L) adapter for JTBD node type detection.

Detector Type

job_trigger

Task Description

Binary classification: given a consumer utterance from a JTBD interview, determine whether it describes a trigger event — something that initiated or prompted the person to seek a solution. Output 'yes' or 'no'.

Base Model

mistralai/Mistral-7B-Instruct-v0.2

Generation Method

Generated using Sakana AI's Text-to-LoRA hypernetwork:

Paper: ICML 2025
Code: https://github.com/SakanaAI/text-to-lora

This adapter was created from a task description alone (zero-shot, no labeled examples).

Configuration

LoRA rank (r): 8
LoRA alpha: 16
Target modules: ['q_proj', 'v_proj']

Usage

python

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

# Load base model
model = AutoModelForCausalLM.from_pretrained(
    "mistralai/Mistral-7B-Instruct-v0.2",
    torch_dtype=torch.bfloat16,
    device_map="auto",
)

# Load T2L adapter
model = PeftModel.from_pretrained(
    model,
    "michaelarutyunov/jtbd-t2l-jobtrigger",
)

# Use for inference
prompt = "<s>[INST] [UTTERANCE]: Your utterance here\nDoes this contain a job_trigger? Answer yes or no.\nAnswer: [/INST]"
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=5)
response = tokenizer.decode(outputs[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True)
print(response)

Integration with D2L

This adapter targets q_proj and v_proj (attention layers). Can be stacked with a D2L adapter (targets down_proj) for combined knowledge:

python

model.load_adapter("path/to/d2l_adapter", adapter_name="d2l")
model.load_adapter("path/to/t2l_job_trigger", adapter_name="t2l")
model.set_adapter(["d2l", "t2l_job_trigger"])

Metadata

Generated: 2026-03-19T20:46:45.591181+00:00
T2L checkpoint: /root/.cache/huggingface/hub/models--SakanaAI--text-to-lora/snapshots/6e571eda2188b216f027263cd28c99c0fdcf2fa3/trained_t2l/mistral_7b_t2l/hypermod.pt
T2L repo commit: 8ba77493eb8c732bac41c22ca76117dcacfcf2a4

Generated as part of Test 2: JTBD Node Type Detection with D2L + T2L