Elevatormusic
/

DeepSeek-V4.1-Flash-1017-2x-DGX-Spark-TensorFold-2x-Faster

Text Generation

tensor-parallel

structured-output

Model card Files Files and versions

DeepSeek-V4.1-Flash-1017-2x-DGX-Spark-TensorFold-2x-Faster

Commit History

Oct 21 20 days ago

16:08
Note that the kit_bench prefill numbers use a small-vocabulary prompt

Elevatormusic

Verified signature
3b9161a

Oct 20 21 days ago

02:27
docs: add an experimental BF16 uncensored option (#1)⁠

Elevatormusic

d44fa51

Oct 18 23 days ago

05:40
Use standard serving terms (prefill, decode, TTFT)

Elevatormusic

Verified signature
db0fa29
05:19
Prefill speeds, 1M context test, memory settings step, fork link

Elevatormusic

Verified signature
51a0ba9
00:44
Card: base model deepseek-ai/DeepSeek-V4.1-Flash-1017, relation quantized

Elevatormusic

Verified signature
e894a7c
00:39
Recipe card: DeepSeek V4.1 Flash 1017 EXL3 on 2x DGX Spark with TensorFold

Elevatormusic

Verified signature
f6a2652
00:39
initial commit

Elevatormusic

Verified signature
59665f7