corpus.blog
56,966 blogs · 4,417,066 posts Claim your blog
Record · corpus.blog/posts/01a0df50-42a2-7315-a56c-0dc7c21cd99d

Rubric-Based Rewards for RL

cameronrwolfe.substack.com · published 16 February 2026

A record is what was published and who pointed at it. The text of the post is not held here: read it at the source, or at its Wayback capture.

Read the post at cameronrwolfe.substack.com Wayback capture
1blogs citing
First cited
22 Feb 2026
Most recent
22 Feb 2026
Rank this month
not ranked
Links from this post
106
Words captured
10,754

Cited by

Blog
In the post
Date
AI for Software Engineers | Logan Thorneloe | Substack
AI’s Biggest Cost Is Cognitive, Not Compute | Weekend Reads 2 Read at the source ↗
22 Feb 2026

Links from this post

Link
Host
https://substackcdn.com/image/fetch/$s_!9S-H!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F9c2a326d-ce08-4c08-b61f-f6729bef3826_2322x1304.png
substackcdn.com
Reward Models Read at the source ↗
cameronrwolfe.substack.com
https://en.wikipedia.org/wiki/BLEU
en.wikipedia.org
https://en.wikipedia.org/wiki/ROUGE_(metric
en.wikipedia.org
https://en.wikipedia.org/wiki/N-gram
en.wikipedia.org
https://arxiv.org/abs/1904.09675
arxiv.org
https://arxiv.org/abs/1707.06875
arxiv.org
https://lmsys.org/blog/2023-03-30-vicuna/
lmsys.org
https://tatsu-lab.github.io/alpaca_eval/
tatsu-lab.github.io
https://lmsys.org/blog/2023-05-03-arena/
lmsys.org
https://lmsys.org/blog/2024-04-19-arena-hard/
lmsys.org
https://substackcdn.com/image/fetch/$s_!zyZu!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F74d136a6-2eb6-4158-8f85-55fa26fa3c8f_1974x1234.png
substackcdn.com
https://en.wikipedia.org/wiki/Likert_scale
en.wikipedia.org
https://cameronrwolfe.substack.com/p/chain-of-thought-prompting-for-llms
cameronrwolfe.substack.com
Using LLM-as-a-Judge For Evaluation: A Complete Guide Read at the source ↗
hamel.dev
https://substackcdn.com/image/fetch/$s_!cVp2!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8e67d161-dacd-48a7-9bb2-eb93052fe583_1724x1268.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!cC5H!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2ccd4de0-4969-4935-ba99-dc91e21e43aa_1450x820.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!6WVC!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fc2d2b494-9f21-46c4-adf5-dbc60bd866ee_2730x1510.png
substackcdn.com
PPO for LLMs: A Guide for Normal People Read at the source ↗
cameronrwolfe.substack.com
Group Relative Policy Optimization (GRPO) Read at the source ↗
cameronrwolfe.substack.com
https://substackcdn.com/image/fetch/$s_!uPv8!,w_1456,c_limit,f_auto,q_auto:good,fl_lossy/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F56eba05c-359c-400d-920f-38a36dd4690a_1920x1078.gif
substackcdn.com
https://substackcdn.com/image/fetch/$s_!zfsl!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffb865992-1eee-4fdb-b98a-165f4d555e11_1774x608.png
substackcdn.com
https://cameronrwolfe.substack.com/p/rlaif-reinforcement-learning-from
cameronrwolfe.substack.com
https://substackcdn.com/image/fetch/$s_!1T_j!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F99f3ffbc-9104-419f-9ccf-3902425a85d8_1580x562.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!xplG!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3f0a1e80-d29b-44e2-b4f3-92803f21a455_2042x1212.png
substackcdn.com
https://arxiv.org/abs/2507.17746
arxiv.org
https://cameronrwolfe.substack.com/i/141159804/different-setups-for-llm-as-a-judge
cameronrwolfe.substack.com
https://substackcdn.com/image/fetch/$s_!9Gs6!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F36d6997c-f2da-44e6-ac55-509f073f6632_2180x1070.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!hVRt!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F1b5392d8-c6c8-4c53-889b-b7ac4b6225ed_1460x566.png
substackcdn.com
https://huggingface.co/datasets/anisha2102/RaR-Medicine
huggingface.co
https://huggingface.co/datasets/anisha2102/RaR-Science
huggingface.co
https://arxiv.org/abs/2502.13124
arxiv.org
https://arxiv.org/abs/2501.15587
arxiv.org
https://huggingface.co/datasets/RJT1990/GeneralThoughtArchive
huggingface.co
https://huggingface.co/Qwen/Qwen2.5-7B
huggingface.co
https://huggingface.co/Qwen/Qwen2.5-7B-Instruct
huggingface.co
https://epoch.ai/benchmarks/gpqa-diamond
epoch.ai
Introducing HealthBench Read at the source ↗
openai.com
https://substackcdn.com/image/fetch/$s_!zcTo!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa57ba595-9591-4539-988c-3a267ab59d87_1592x882.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!Z_4o!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ff252947f-8ea2-48ba-bc42-13ed3031c03a_2060x734.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!48A-!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F209e88f2-4eed-4cea-a016-e0185bc3779c_2050x936.png
substackcdn.com
https://arxiv.org/abs/2508.12790
arxiv.org
Asymmetry of verification and verifier’s rule Read at the source ↗
jasonwei.net
https://substackcdn.com/image/fetch/$s_!JmJ3!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ff9d5109e-ebfd-41fd-b80b-623d7182677d_2326x906.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!tTh5!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F67bef4a7-0f51-4ea2-8c9f-554e9aa0f9d1_1238x1394.png
substackcdn.com
https://cameronrwolfe.substack.com/i/179769076/rlvr-with-grpo
cameronrwolfe.substack.com
https://substackcdn.com/image/fetch/$s_!qXlF!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F1d06eba0-dd0b-4748-b670-f48295e4cc6c_1278x1018.png
substackcdn.com
https://huggingface.co/Qwen/Qwen3-30B-A3B
huggingface.co
https://arxiv.org/abs/2412.19437
arxiv.org
https://substackcdn.com/image/fetch/$s_!MmTq!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7a354560-7f5f-4f75-b614-afb34ecc3894_1288x306.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!CPpf!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2941e56b-30c0-4ce1-9558-f7d8e92323e5_1270x206.png
substackcdn.com
Continual Learning with RL for LLMs Read at the source ↗
cameronrwolfe.substack.com
https://arxiv.org/abs/2510.07743
arxiv.org
https://huggingface.co/datasets/OpenRubrics/OpenRubrics
huggingface.co
https://huggingface.co/datasets/openbmb/UltraFeedback
huggingface.co
https://huggingface.co/datasets/MegaScience/MegaScience
huggingface.co
https://huggingface.co/datasets/FreedomIntelligence/medical-o1-reasoning-SFT
huggingface.co
https://arxiv.org/abs/2411.15124
arxiv.org
https://substackcdn.com/image/fetch/$s_!xwPa!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F08c12e30-3151-4d3c-b642-ec4bbba84625_2386x726.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!U2Ni!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Faf55c52b-1691-4738-b306-c5a019a92acb_1564x895.png
substackcdn.com
https://huggingface.co/Qwen/Qwen3-4B
huggingface.co
https://huggingface.co/Qwen/Qwen3-8B
huggingface.co
https://substackcdn.com/image/fetch/$s_!pSGK!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F13911e95-b761-4a5f-8695-3e28d00ef417_1582x812.png
substackcdn.com
https://arxiv.org/abs/2311.07911
arxiv.org
https://arxiv.org/abs/2401.03601
arxiv.org
https://arxiv.org/abs/2507.02833
arxiv.org
https://substackcdn.com/image/fetch/$s_!ob_C!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffba83fba-5066-40a3-9d65-17628483294a_1578x722.png
substackcdn.com
https://arxiv.org/abs/2511.19399
arxiv.org
https://blog.google/products-and-platforms/products/gemini/google-gemini-deep-research/
blog.google
Deep Research Read at the source ↗
openai.com
https://www.anthropic.com/engineering/multi-agent-research-system
anthropic.com
https://substackcdn.com/image/fetch/$s_!8OAX!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F36d75c42-fc6d-4268-9678-3f28532f3bef_2018x1476.png
substackcdn.com
https://cameronrwolfe.substack.com/p/teaching-language-models-to-use-tools
cameronrwolfe.substack.com
https://substackcdn.com/image/fetch/$s_!D4L9!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fe2189065-9373-46ac-a9d7-4e9ab566a57f_2266x636.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!rj--!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F793e6538-cd2e-4e52-8fd6-d41fd544c6af_2394x1120.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!qMID!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6a24602f-f704-4f85-889a-fe212299727f_1966x950.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!iVsF!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F74b511c4-0ca4-4892-9ab7-4c075a0b52d9_2148x1406.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!sLQ4!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcb503197-6f95-40a8-9583-8b18b9891c95_2380x898.png
substackcdn.com
https://huggingface.co/datasets/rl-research/dr-tulu-sft-data
huggingface.co
https://cameronrwolfe.substack.com/i/181791956/dapo-an-open-source-llm-reinforcement-learning-system-at-scale-1
cameronrwolfe.substack.com
https://huggingface.co/spaces/HuggingFaceTB/smol-training-playbook
huggingface.co
https://substackcdn.com/image/fetch/$s_!hAdM!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2435b96a-7d03-4623-b862-79afaa5862c3_2141x805.png
substackcdn.com
https://arxiv.org/abs/2504.10861
arxiv.org
https://arxiv.org/abs/2509.00496
arxiv.org
https://arxiv.org/abs/2506.11763
arxiv.org
https://huggingface.co/Alibaba-NLP/Tongyi-DeepResearch-30B-A3B
huggingface.co
https://arxiv.org/abs/2602.01511
arxiv.org
https://substackcdn.com/image/fetch/$s_!4zwp!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fe396bca7-e97e-40e9-8332-ba003b1e2d39_2505x1405.png
substackcdn.com
https://arxiv.org/abs/2310.01377
arxiv.org
https://arxiv.org/abs/2506.20737
arxiv.org
Direct Preference Optimization (DPO) Read at the source ↗
cameronrwolfe.substack.com
https://cameronrwolfe.substack.com/i/169926007/direct-alignment-techniques
cameronrwolfe.substack.com
https://substackcdn.com/image/fetch/$s_!_coW!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F1d0bf86d-09aa-41c3-b88e-79f0368c8d26_1243x1261.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!7dES!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F5bd1c7d2-0698-4ba3-9b69-ff7e2c9bcc2b_2280x1336.png
substackcdn.com
https://writingpreferencebench.github.io/
writingpreferencebench.github.io
https://substackcdn.com/image/fetch/$s_!7wvl!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb25d8ac1-08f8-434c-8bf4-d2fb72c92e16_1532x844.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!cxW-!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fadaa6c51-0537-46cb-8c2f-232f2b30cea5_2472x1170.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!93bd!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F40f57385-3aef-4efc-8d9b-37f83c89a29c_1710x653.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!JSND!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F9a0213a4-0890-4026-afa7-6e18be32f74d_1064x1520.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!ktZr!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F0c53199e-bb4e-4384-a1ec-0766622dfcf9_1848x1112.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!Fjct!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa1fa7955-0932-4f88-861c-f54cf1afe289_2006x1322.png
substackcdn.com
https://rlhfbook.com/c/09-rejection-sampling
rlhfbook.com
https://cameronrwolfe.substack.com/i/166169560/how-do-rms-work
cameronrwolfe.substack.com
https://github.com/rlresearch/dr-tulu
github.com
https://huggingface.co/collections/rl-research/dr-tulu
huggingface.co
Olmo 3 and the Open LLM Renaissance Read at the source ↗
cameronrwolfe.substack.com