[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

@BlinkDL_AI "RWKV-7 G1a 1.5B : pushing RNN further with better data (= G1 +1T tokens) supports two reasoning styles and length control. Download: 🚀 and G1a 0.4/0.1B & G0a 7B updated G0 13B XX% trained (already XXXX% MMLU 🙂)"
X Link @BlinkDL_AI 2025-09-23T13:29Z 9195 followers, 19.5K engagements

"RWKV-7 G1a 2.9B more evals: MMLU Pro XX% (+CoT) GSM8K XX% MATH 50%. Note this is a base model no mid-training no post-training. I just add everything to pretraining dataset"
X Link @BlinkDL_AI 2025-10-09T14:14Z 9193 followers, 3891 engagements

"@GusarichOnX No that's wrong 🙂 Please try with"
X Link @BlinkDL_AI 2025-10-14T03:49Z 9192 followers, XX engagements

"@inductionheads The EmbROSASampling part is neurosymbolic and now trainable by perturbing symbols 🙂"
X Link @BlinkDL_AI 2025-10-15T11:39Z 9191 followers, XXX engagements

"RWKV-7 G1a 2.9B : pure RNN surpassing Gemma3 4B and Llama3.2 3B in some areas supports two reasoning styles and length control. Download: 🚀 and G1a 1.5/0.4/0.1B & G0a 7B updated G0 13B release in Oct"
X Link @BlinkDL_AI 2025-09-26T15:13Z 9192 followers, 44.7K engagements

"RWKV8 ROSA 🌹 simply scales producing mysterious new languages. Training small LMs soon 🙂 Code:"
X Link @BlinkDL_AI 2025-10-17T17:26Z 9195 followers, 9067 engagements