1. A genuine willingness and excitement to learn new technologies, even if they fall outside the current tech stack.

        1. randy@mxgs76:~ Current Development Stack…...local machine

$ uname -a

Linux mxgs76 6.4.0-1mx-ahs-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.4.4-1~mx23+1 (2023-07-26) x86_64 GNU/Linux

        1. Current Client technical stack: Oracle v19.x , MSFT SQL ver ….

        2. telecon 17Nov’23, 9:40am to 10:10am CST.

        3. Chain-of-Thought(CoT); Measuring Faithfulness in Chain-of-Thought Reasoningi.

            1. Open-Source, OpenAI’s Q star, Q*.

        4. AI tools such as ChatGPT can best be incorporated within software

        5. related to the Oil & Gas Drilling and Production industry.

            1. End Client PoC(ECPoC) define CoT.

            2. Open source github source code.

            3. ECPoC domain name , private, not accessible outside ECPoC.




            7. https://

            8. https://ecpoc.test



            11. ( _ _ _ _ _ _ _ )ecpoc.(xyz)

            12. AI cloud; ECPoC domain name , private,

              not accessible outside ECPoC.

        1. Flare gas powered data center.

Randy Middleton

Champs Construction LLC, sole proprietor, 2007 to present 

 315 Prairie Creek Trail, Murphy, TX 75094;home address,

281 736-1696 ,

arabic58 at GitHub,

 “WhatsApp contact”;

 Linkedin ,

Telegram @arabic58 

iMeasuring Faithfulness in Chain-of-Thought Reasoning;


Large language models (LLMs) perform better when they produce step-by-step, “Chain-ofThought” (CoT) reasoning before answering a question, but it is unclear if the stated reasoning is a faithful explanation of the model’s actual reasoning (i.e., its process for answering the question). We investigate hypotheses for how CoT reasoning may be unfaithful, by examining how the model predictions change when we intervene on the CoT (e.g., by adding mistakes or paraphrasing it). Models show large variation across tasks in how strongly they condition on the CoT when predicting their answer, sometimes relying heavily on the CoT and other times primarily ignoring it. CoT’s performance boost does not seem to

come from CoT’s added test-time compute alone or from information encoded via the particular

phrasing of the CoT. As models become larger and more capable, they produce less faithful reasoning on most tasks we study. Overall, our results suggest that CoT can be faithful if the circumstances such as the model size and task are carefully chosen.

Tamera Lanham

Anna Chen Ansh Radhakrishnan Benoit Steiner Carson Denison Danny Hernandez Dustin Li Esin Durmus

Evan Hubinger Jackson Kernion Kamile Lukosiute Karina Nguyen Newton Cheng Nicholas Joseph

Nicholas Schiefer Oliver Rausch Robin Larson Sam McCandlish Sandipan Kundu Saurav Kadavath

Shannon Yang Thomas Henighan Timothy Maxwell Timothy Telleen-Lawton Tristan Hume

Zac Hatfield-Dodds

Jared Kaplan Jan Brauner Samuel R. Bowman Ethan Perez;

[ All authors at Anthropic, except Jan Brauner who is at University of Oxford. Correspondence to: Tamera Lanham < >, Ethan Perez < >.]

© Anthropic , 2023-July-18.