Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO-MTP-DFlash2

He aha Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO-MTP-DFlash2

He tauira 27B-tawhiwhi, whakakaha-a-tuhinga me te whakaaro nui, e hāngai ana ki Qwen, i arotahi mā te whakangungu i ngā wāhanga maha tae atu ki SFT, RLOO me MTP, e arotahi ana ki te whakaaro pai, te hōtaka me ngā mahi whakaaro uaua.

Ngā āhuatanga matua

• I arotahi mā te whakangungu i ngā wāhanga maha tae atu ki SFT, RLOO me MTP e hāngai ana ki Qwen3.8-27B
• Ka tino whakaiti i te raru whakawhāiti 94K i puta mai i te whakatauira tauira tukurua
• E tautoko ana i ngā whakahōputu whakarawe maha tae atu ki BF16, FP8, NVFP4, INT4, INT8 me GGUF
• Tautoko māori mō te kuputuhi + whakaahua + ataata whakaurunga multimodal
• Kua whakaurua ngā tikanga whakatere whakatauira whakapae MTP me DFlash2 e tūpono ana

Mā wai

Ka whai pānga ngā kaiwhakawhanake me ngā kairangahau e hiahia ana ki te hōtaka pai, ngā mahi whakaaro uaua me ngā mahi multimodal, e tika ana mō ngā āhuatanga me ngā whakaritenga teitei mō te mana roa o te putanga tauira me te pūmautanga whakatauira.

Me pēhea te tīmata

1. Toro atu ki https://huggingface.co/nerkyor/Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO-MTP-DFlash2
2. Tikiake i ngā taumaha safetensors i roto i te whakahōputu e hiahiatia ana (BF16, FP8, NVFP4, INT4, GGUF, aha noa)
3. Tukuna te tauira tika ki SGLang, vLLM rānei (kua whakamātauria te putanga FP8, e mahi ana ki runga i vLLM)

I whakahiatohia mai i ngā whakaahuatanga tūmatanui i te pae mana / README / whārangi whakarewatanga o te taputapu. Kāore anō kia whakamanahia ngā utu me ngā tau kaiwhakamahi; tirohia te pae mana. I hangaia ētahi wāhanga mā te āwhina AI. Kua tuhia ki te reo Hainamana me te reo Ingarihi; ko ētahi atu reo he whakamāoritanga mihini.