Avi Fenesh
Avifenesh
AI & ML interests
Inference and quant research, fine-tuning and specialist models, BW, self-hosting, on-prem, and intelligence open for everyone
Recent Activity
updated a model about 16 hours ago
tiyuvta/MiMo-V2.6-Flash-RL-NVFP4 liked a model 2 days ago
tiyuvta/MiMo-V2.6-Flash-RL-NVFP4 published a model 2 days ago
tiyuvta/MiMo-V2.6-Flash-RL-NVFP4Organizations
Lowest-latency hosted provider for DeepSeek V4.1 Flash from Northern Virginia (US East)?
3
#57 opened 8 days ago
by 
KekeJ
Any chance of uploading Ornith1.5-9B-NVFP4-MTP-GGUF
7
#1 opened about 1 month ago
by 
sa13ma
Context section correction: 1M demonstrated, chunk-schedule fix, honest boundaries
#4 opened 25 days ago
by 
Avifenesh
Context section correction
#3 opened 25 days ago
by Avifenesh
Context section correction
#2 opened 25 days ago
by Avifenesh
docs: attribute the hosted Ornith link
1
#3 opened 27 days ago
by Avifenesh
docs: link the hosted Ornith API
1
#2 opened 27 days ago
by Avifenesh
docs: link the hosted Qwen3.8 API
1
#3 opened 27 days ago
by Avifenesh
No speculative decoding, No QAT
8
#21 opened about 1 month ago
by 
engrtipusultan
DFlash2?
👀 5
1
#79 opened about 1 month ago
by 
guarism0
NVFP4/MXFP4 GGUF Version?
👍 3
2
#75 opened about 1 month ago
by 
CYISNOTHERE
Unsloth GGUF's please
1
#3 opened about 1 month ago
by 
jlindeberg
MTP Support
🚀 9
2
#2 opened about 1 month ago
by 
DefaultDF
MTP (SPEC_DRAFT_N_MAX) leads to Nvidia RC Watchdog killing the kernel, llama.cpp calls SIGABRT
➕ 1
5
#122 opened about 1 month ago
by 
Nosepicker3
Small Quants with integrated MTP dont make sense
👍 5
4
#37 opened about 1 month ago
by 
TheWegemann
qwen3.8-27b on one pro 6000: 140 tok/s, day after release
7
#101 opened about 1 month ago
by Avifenesh
MTP speculation is net-negative under Ollama's GGUF path, strongly positive under MLX (measured on M4 Pro)
5
#80 opened about 1 month ago
by 
OtsoaD