All news

Ttogether_ai_blog08:00First-partyAnalysis

AI score66

English translation is being prepared.

#Evaluations / benchmarks#English text is being prepared.#English text is being prepared.

Ttogether_ai_blog08:00First-partyAnalysis

AI score75

English translation is being prepared.

#Evaluations / benchmarks#English text is being prepared.#English text is being prepared.

Nnvidia_developer00:00First-partyAnalysis

AI score54

English translation is being prepared.

#Tutorials / practice#Embodied AI#English text is being prepared.

Aarmin_ronacher_ai08:00Analysis

AI score63

English translation is being prepared.

#Tutorials / practice#Reasoning#OpenAI

Hhugging_face08:00First-partyAnalysis

AI score60

English translation is being prepared.

#Tutorials / practice#RAG#Open-source ecosystem

Ttogether_ai_blog08:00First-partyAnalysis

AI score71

English translation is being prepared.

#Evaluations / benchmarks#English text is being prepared.#Agents

Llmsys_blog08:00First-partyAnalysis

AI score51

English translation is being prepared.

#Tutorials / practice#English text is being prepared.#Open-source ecosystem

Aahead_of_ai19:54Analysis

AI score56

English translation is being prepared.

#Tutorials / practice#English text is being prepared.#English text is being prepared.

Aanthropic_news03:16First-partyAnalysis

AI score75

English translation is being prepared.

#English text is being prepared.#Anthropic#Google

AAnt Ling: Developer Blog08:00First-partyAnalysis

AI score58

Ant Ling closes the single-machine Agentic RL loop with Ling-3.0-tiny and AReno

Ant Ling has published a practice article describing how it used the local post-training toolkit AReno on a DGX Spark to perform Agentic RL on Ling-3.0-tiny, using tic-tac-toe as a minimal validation task to complete the four-step loop of baseline evaluation, Reward definition, post-training, and re-evaluation in the same environment. The experiment used GSPO for 400 training steps; rollout/rewards_mean rose from about -0.5 to around 0.4, response_len dropped to about 850 tokens, and tool calling and move selection became more stable in the re-evaluation. Ling-3.0-tiny is a hybrid reasoning model with 7.9B total parameters and 1.3B activated parameters; BF16, FP8, and INT4 versions have been open-sourced, and AReno provides a Docker reproduction tutorial.

#Tutorials / practice#Agents#Open-source ecosystem

Hhugging_face01:16First-partyAnalysis

AI score64

English translation is being prepared.

#Tutorials / practice#Agents#Embodied AI

Ddeepseek_ai21:02

AI score74

English translation is being prepared.

#English text is being prepared.#Agents#English text is being prepared.

Oopenrouter_announcements08:00First-partyAnalysis

AI score51

English translation is being prepared.

#Evaluations / benchmarks#Search#Agents

Llinear_now20:19First-partyAnalysis

AI score73

English translation is being prepared.

#Tutorials / practice#Agents#English text is being prepared.

Tthe_register00:44Analysis

AI score60

English translation is being prepared.

#Safety / alignment#Agents#English text is being prepared.

Ttogether_ai_blog08:00First-partyAnalysis

AI score67

English translation is being prepared.

#Evaluations / benchmarks#English text is being prepared.#Agents

Nnvidia_developer00:00First-partyAnalysis

AI score62

English translation is being prepared.

#Papers/Research#Embodied AI#Multimodal

Mmeituan_longcat22:25

AI score35

English translation is being prepared.

#Product updates#Agents#English text is being prepared.

Mmeituan_longcat22:25Analysis

AI score29

English translation is being prepared.

#Product updates#Video#English text is being prepared.

Kkarpathy00:10

AI score48

English translation is being prepared.

#Evaluations / benchmarks#Video#Open-source ecosystem