Google Gemma 4 models (E2B, E4B) in MNN Q4 format for TokForge mobile inference. Requires TokForge 3.4.9+.
Ivan M
darkmaniac7
AI & ML interests
None yet
Recent Activity
updated a model 1 day ago
darkmaniac7/tokforge-assets updated a model 1 day ago
darkmaniac7/Huihui-gemma-4-E2B-it-abliterated-MNN updated a model 1 day ago
darkmaniac7/Gemma-4-E2B-it-heretic-ara-MNNOrganizations
None yet
Uncensored Llama/Dolphin/Mistral models in MNN Q4 for TokForge mobile inference.
-
darkmaniac7/Panth-L3-Blackroot-Nephra-8B-MNN
Text Generation • Updated • 208 -
darkmaniac7/Dolphin3.0-Llama3.1-8B-MNN
Text Generation • Updated • 191 -
darkmaniac7/Meta-Llama-3.1-8B-Instruct-abliterated-MNN
Text Generation • Updated • 192 -
darkmaniac7/DarkIdol-Llama-3.1-8B-Instruct-1.2-Uncensored-MNN
Text Generation • Updated • 182
Small MNN draft models and speculative-decoding bundles for TokForge on Android. Includes practical Qwen3 0.6B drafts plus experimental variants.
Top open-source models in MNN Q4 for TokForge mobile inference. Google, Meta, Mistral, DeepSeek, HuggingFace.
Uncensored Qwen3/3.5 models in MNN Q4 format for TokForge mobile inference. Josiefied, Heretic, and Claude variants.
-
darkmaniac7/Josiefied-Qwen3.5-0.8B-gabliterated-v1-MNN
Text Generation • Updated • 339 • 2 -
darkmaniac7/Josiefied-Qwen3-4B-abliterated-v2-MNN
Text Generation • Updated • 494 -
darkmaniac7/Josiefied-Qwen3-8B-abliterated-v1-MNN
Text Generation • Updated • 554 • 1 -
darkmaniac7/Josiefied-Qwen3-14B-abliterated-v3-MNN
Text Generation • Updated • 326
Google Gemma 4 models (E2B, E4B) in MNN Q4 format for TokForge mobile inference. Requires TokForge 3.4.9+.
Top open-source models in MNN Q4 for TokForge mobile inference. Google, Meta, Mistral, DeepSeek, HuggingFace.
Uncensored Llama/Dolphin/Mistral models in MNN Q4 for TokForge mobile inference.
-
darkmaniac7/Panth-L3-Blackroot-Nephra-8B-MNN
Text Generation • Updated • 208 -
darkmaniac7/Dolphin3.0-Llama3.1-8B-MNN
Text Generation • Updated • 191 -
darkmaniac7/Meta-Llama-3.1-8B-Instruct-abliterated-MNN
Text Generation • Updated • 192 -
darkmaniac7/DarkIdol-Llama-3.1-8B-Instruct-1.2-Uncensored-MNN
Text Generation • Updated • 182
Uncensored Qwen3/3.5 models in MNN Q4 format for TokForge mobile inference. Josiefied, Heretic, and Claude variants.
-
darkmaniac7/Josiefied-Qwen3.5-0.8B-gabliterated-v1-MNN
Text Generation • Updated • 339 • 2 -
darkmaniac7/Josiefied-Qwen3-4B-abliterated-v2-MNN
Text Generation • Updated • 494 -
darkmaniac7/Josiefied-Qwen3-8B-abliterated-v1-MNN
Text Generation • Updated • 554 • 1 -
darkmaniac7/Josiefied-Qwen3-14B-abliterated-v3-MNN
Text Generation • Updated • 326
Small MNN draft models and speculative-decoding bundles for TokForge on Android. Includes practical Qwen3 0.6B drafts plus experimental variants.