OPEx: A Large Language Model-Powered Framework for Embodied Instruction Following
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Embodied Instruction Following (EIF) is crucial for understanding natural language in a practical context, requiring agents to follow verbal instructions for complex tasks. Traditionally, EIF relies heavily on expert annotations for learning, which are costly and sometimes unattainable. Recent research shows Large Language Models (LLMs) can use their reasoning ability to help in EIF with minimal examples, but applying LLMs directly faces issues like hallucinations and partially observable environment. To bridge the gap, we introduce OPEx, a new LLM-based method for EIF that needs far less specific data. OPEx uses three LLMs for different roles: observing to gather environment data, planning by breaking down instructions, and executing tasks with learned skills. Our tests reveal OPEx significantly outperforms the FILM baseline, with 90% less training data for planning tasks and achieving up to 38% performance gain when FILM is trained on identical data.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it