Experimental Machines

non-profit

https://experimentalmachines.org/

ExperimentalMachines

experimentalmachines

Activity Feed Request to join this org

AI & ML interests

Accelerated Computing

Recent Activity

alpharomercoma updated a Space about 3 hours ago

experimentalmachines/README

alpharomercoma updated a model 1 day ago

experimentalmachines/LFM2.5-2.6B-heretic-ExecuTorch

alpharomercoma updated a model 1 day ago

experimentalmachines/LFM2.5-1.2B-Instruct-heretic-ExecuTorch

View all activity

experimentalmachines 's collections 7

Qwen3 for ExecuTorch

Alibaba's Qwen3 chat models for ExecuTorch XNNPACK. 8da4w int4; the 1.7B is GPTQ-solved, the rest round-to-nearest.

experimentalmachines/Qwen3-0.6B-ExecuTorch

Text Generation • Updated 3 days ago • 63 • 2
experimentalmachines/Qwen3-1.7B-ExecuTorch

Text Generation • Updated 1 day ago • 42
experimentalmachines/Qwen3-4B-ExecuTorch

Text Generation • Updated 8 days ago • 37
experimentalmachines/Qwen3-4B-Instruct-2507-ExecuTorch

Text Generation • Updated 8 days ago • 21

LFM2.5 Abliterated for ExecuTorch

Liquid AI's LFM2.5 with refusals removed by heretic, for ExecuTorch XNNPACK. 8da4w int4, GPTQ-solved. Five windows, 2k-32k.

experimentalmachines/LFM2.5-1.2B-Instruct-heretic-ExecuTorch

Text Generation • Updated 1 day ago • 27
experimentalmachines/LFM2.5-2.6B-heretic-ExecuTorch

Text Generation • Updated 1 day ago • 27

Llama 3.2 for ExecuTorch

Meta's Llama 3.2 instruct models for ExecuTorch XNNPACK. 8da4w int4, round-to-nearest. One .pte per context window.

experimentalmachines/Llama-3.2-1B-Instruct-ExecuTorch

Text Generation • Updated 8 days ago • 24
experimentalmachines/Llama-3.2-3B-Instruct-ExecuTorch

Text Generation • Updated 8 days ago • 83

QwenGrad for ExecuTorch

OpenGrad's Qwen3.5-2B M1-DPO-v2 for ExecuTorch. Built by hand rather than the pipeline: no export report and no decision gate.

experimentalmachines/QwenGrad-Qwen3.5-2B-M1-DPO-v2-ExecuTorch-CPU

Updated 9 days ago • 9

Qwen2.5 for ExecuTorch

Alibaba's Qwen2.5 chat models for ExecuTorch XNNPACK. 8da4w int4, round-to-nearest. One .pte per context window.

experimentalmachines/Qwen2.5-0.5B-Instruct-ExecuTorch

Text Generation • Updated 8 days ago • 23
experimentalmachines/Qwen2.5-1.5B-Instruct-ExecuTorch

Text Generation • Updated 8 days ago • 21
experimentalmachines/Qwen2.5-3B-Instruct-ExecuTorch

Text Generation • Updated 8 days ago • 21
experimentalmachines/Qwen2.5-Math-1.5B-Instruct-ExecuTorch

Text Generation • Updated 8 days ago • 15

SmolLM2 for ExecuTorch

Hugging Face's SmolLM2 instruct models for ExecuTorch XNNPACK. 8da4w int4, round-to-nearest. Measured degraded; writes no tool calls.

experimentalmachines/SmolLM2-135M-Instruct-ExecuTorch

Text Generation • Updated 3 days ago • 49
experimentalmachines/SmolLM2-360M-Instruct-ExecuTorch

Text Generation • Updated 8 days ago • 27

LFM2.5 for ExecuTorch

Liquid AI's LFM2.5 instruct models for ExecuTorch XNNPACK. 8da4w int4, GPTQ-solved. Five windows, 2k-32k. Abliterated ones in their own collection.

experimentalmachines/LFM2.5-1.2B-Instruct-ExecuTorch

Text Generation • Updated 1 day ago • 118
experimentalmachines/LFM2.5-2.6B-ExecuTorch

Text Generation • Updated 1 day ago • 448

Qwen3 for ExecuTorch

Alibaba's Qwen3 chat models for ExecuTorch XNNPACK. 8da4w int4; the 1.7B is GPTQ-solved, the rest round-to-nearest.

experimentalmachines/Qwen3-0.6B-ExecuTorch

Text Generation • Updated 3 days ago • 63 • 2
experimentalmachines/Qwen3-1.7B-ExecuTorch

Text Generation • Updated 1 day ago • 42
experimentalmachines/Qwen3-4B-ExecuTorch

Text Generation • Updated 8 days ago • 37
experimentalmachines/Qwen3-4B-Instruct-2507-ExecuTorch

Text Generation • Updated 8 days ago • 21

Qwen2.5 for ExecuTorch

Alibaba's Qwen2.5 chat models for ExecuTorch XNNPACK. 8da4w int4, round-to-nearest. One .pte per context window.

experimentalmachines/Qwen2.5-0.5B-Instruct-ExecuTorch

Text Generation • Updated 8 days ago • 23
experimentalmachines/Qwen2.5-1.5B-Instruct-ExecuTorch

Text Generation • Updated 8 days ago • 21
experimentalmachines/Qwen2.5-3B-Instruct-ExecuTorch

Text Generation • Updated 8 days ago • 21
experimentalmachines/Qwen2.5-Math-1.5B-Instruct-ExecuTorch

Text Generation • Updated 8 days ago • 15

LFM2.5 Abliterated for ExecuTorch

Liquid AI's LFM2.5 with refusals removed by heretic, for ExecuTorch XNNPACK. 8da4w int4, GPTQ-solved. Five windows, 2k-32k.

experimentalmachines/LFM2.5-1.2B-Instruct-heretic-ExecuTorch

Text Generation • Updated 1 day ago • 27
experimentalmachines/LFM2.5-2.6B-heretic-ExecuTorch

Text Generation • Updated 1 day ago • 27

SmolLM2 for ExecuTorch

Hugging Face's SmolLM2 instruct models for ExecuTorch XNNPACK. 8da4w int4, round-to-nearest. Measured degraded; writes no tool calls.

experimentalmachines/SmolLM2-135M-Instruct-ExecuTorch

Text Generation • Updated 3 days ago • 49
experimentalmachines/SmolLM2-360M-Instruct-ExecuTorch

Text Generation • Updated 8 days ago • 27

Llama 3.2 for ExecuTorch

Meta's Llama 3.2 instruct models for ExecuTorch XNNPACK. 8da4w int4, round-to-nearest. One .pte per context window.

experimentalmachines/Llama-3.2-1B-Instruct-ExecuTorch

Text Generation • Updated 8 days ago • 24
experimentalmachines/Llama-3.2-3B-Instruct-ExecuTorch

Text Generation • Updated 8 days ago • 83

LFM2.5 for ExecuTorch

Liquid AI's LFM2.5 instruct models for ExecuTorch XNNPACK. 8da4w int4, GPTQ-solved. Five windows, 2k-32k. Abliterated ones in their own collection.

experimentalmachines/LFM2.5-1.2B-Instruct-ExecuTorch

Text Generation • Updated 1 day ago • 118
experimentalmachines/LFM2.5-2.6B-ExecuTorch

Text Generation • Updated 1 day ago • 448

QwenGrad for ExecuTorch

OpenGrad's Qwen3.5-2B M1-DPO-v2 for ExecuTorch. Built by hand rather than the pipeline: no export report and no decision gate.

experimentalmachines/QwenGrad-Qwen3.5-2B-M1-DPO-v2-ExecuTorch-CPU

Updated 9 days ago • 9

AI & ML interests

Recent Activity

Team members 4

experimentalmachines 's collections 7