--- license: unknown --- This is my "F***ing New Guy" quants for the merges I've been playing with. "I don't know what I'm doing" is an understatement, but until I find an obi-wan, I'm just fumbling around in the dark. Some of the stuff is kinda useable though. Specifically my Stage2-12B seemed hyperfocused on single-persona roles and Stage1 has some weird lightning in a bottle element to it that I can't quite articulate, but it's worth a nudge if you don't mind a little instability. As far as I'm aware theyre all creative commons 4 copyright, but I'll be honest, I was definitely making some of these at 2 am. The quants are only available in imatrix IQ4_XS as that's what I use for testing the temperment of the model since it is quick and easy to make and use on my 6900xt. Even with public listings being free - I probably won't exacerbate the storage issue and will purge any HF models that don't meet muster. When I get one that I really like I'll go the whole nine yards on quant variety, until then, these are concept cars. Some hot glue is expected. The following YAML configuration was used to produce these models: Stage1: (Interesting almost authentic responses in RP, has some quirks and for the love of god don't let it use ♡ without editing it out or it spreads and makes a chat unusable) ```yaml models: - model: anthracite-org/magnum-v4-12b parameters: density: 0.3 weight: 0.3 - model: TheDrummer/Rocinante-12B-v1.1 parameters: density: 0.7 weight: 0.5 - model: VongolaChouko/Starcannon-Unleashed-12B-v1.0 parameters: density: 0.1 weight: 0.2 merge_method: ties base_model: TheDrummer/Rocinante-12B-v1.1 parameters: normalize: true int8_mask: true dtype: bfloat16 ``` Stage2: (Over most models I've used, this one had a large swing toward focusing on its own RP and it was difficult to get it to even acknowledge implied characters - decent 1v1) ```yaml base_model: TheDrummer/Rocinante-12B-v1.1 dtype: bfloat16 merge_method: della_linear parameters: epsilon: 0.05 int8_mask: 1.0 lambda: 0.8 slices: - sources: - layer_range: [0, 40] model: anthracite-org/magnum-v4-12b parameters: density: 0.7 weight: 0.25 - layer_range: [0, 40] model: nbeerbower/Lyra4-Gutenberg-12B parameters: density: 0.6 weight: 0.2 - layer_range: [0, 40] model: MarinaraSpaghetti/NemoMix-Unleashed-12B parameters: density: 0.4 weight: 0.15 - layer_range: [0, 40] model: TheDrummer/Rocinante-12B-v1.1 parameters: density: 0.5 weight: 0.4 ``` Stage3: (TBD - initial vibes were rocinante but worse) ```yaml base_model: TheDrummer/Rocinante-12B-v1.1 dtype: bfloat16 merge_method: della_linear parameters: epsilon: 0.05 int8_mask: 1.0 lambda: 0.8 slices: - sources: - layer_range: [0, 40] model: anthracite-org/magnum-v4-12b parameters: density: 0.7 weight: 0.20 - layer_range: [0, 40] model: nbeerbower/Lyra4-Gutenberg-12B parameters: density: 0.7 weight: 0.25 - layer_range: [0, 40] model: MarinaraSpaghetti/NemoMix-Unleashed-12B parameters: density: 0.4 weight: 0.15 - layer_range: [0, 40] model: TheDrummer/Rocinante-12B-v1.1 parameters: density: 0.5 weight: 0.4 ``` Stage4: (TBD - Stage 3 samesame) ```yaml base_model: TheDrummer/Rocinante-12B-v1.1 dtype: bfloat16 merge_method: della_linear parameters: epsilon: 0.05 int8_mask: 1.0 lambda: 0.7 slices: - sources: - layer_range: [0, 40] model: anthracite-org/magnum-v4-12b parameters: density: 0.3 weight: 0.15 - layer_range: [0, 40] model: nbeerbower/Lyra4-Gutenberg-12B parameters: density: 0.8 weight: 0.3 - layer_range: [0, 40] model: MarinaraSpaghetti/NemoMix-Unleashed-12B parameters: density: 0.3 weight: 0.1 - layer_range: [0, 40] model: nothingiisreal/MN-12B-Starcannon-v3 parameters: density: 0.5 weight: 0.1 - layer_range: [0, 40] model: TheDrummer/Rocinante-12B-v1.1 parameters: density: 0.5 weight: 0.35 ``` Stage5: (Failure - Language felt generally unremarkable versus stage 1, lyra, gutenberg or dare has introduced 1st/3rd party switching mid-chat) ```yaml models: - model: anthracite-org/magnum-v4-12b parameters: density: 0.6 weight: 0.3 - model: nbeerbower/Lyra4-Gutenberg-12B parameters: density: 0.83 weight: 0.4 merge_method: dare_ties base_model: TheDrummer/Rocinante-12B-v1.1 parameters: normalize: true int8_mask: true dtype: bfloat16 ``` Stage6: (Absolute failure - Yandere simulator, switches 1st/3rd often - not sure if gutenberg training again or chronos as a new addition - might be moist if not so unhinged) ```yaml models: - model: anthracite-org/magnum-v4-12b parameters: density: 0.69 weight: 0.12 - model: elinas/Chronos-Gold-12B-1.0 parameters: density: 0.69 weight: 0.25 - model: MarinaraSpaghetti/NemoMix-Unleashed-12B parameters: density: 0.42 weight: 0.11 - model: TheDrummer/Rocinante-12B-v1.1 parameters: density: 0.42 weight: 0.24 merge_method: dare_ties base_model: nbeerbower/mistral-nemo-gutenberg-12B parameters: normalize: true int8_mask: true dtype: bfloat16 ``` Stage7: (TBD, initially decent, if a tad forward) ```yaml models: - model: anthracite-org/magnum-v4-12b parameters: density: 0.69 weight: 0.22 - model: elinas/Chronos-Gold-12B-1.0 parameters: density: 0.42 weight: 0.10 - model: MarinaraSpaghetti/NemoMix-Unleashed-12B parameters: density: 0.42 weight: 0.13 - model: TheDrummer/Rocinante-12B-v1.1 parameters: density: 0.69 weight: 0.24 merge_method: ties base_model: nbeerbower/mistral-nemo-bophades-12B parameters: normalize: true int8_mask: true dtype: bfloat16 ``` Stage8: (TBD, initially decent, some issues with situ awareness) ```yaml models: - model: anthracite-org/magnum-v4-12b parameters: density: 0.69 weight: 0.29 - model: MarinaraSpaghetti/NemoMix-Unleashed-12B parameters: density: 0.42 weight: 0.12 - model: TheDrummer/Rocinante-12B-v1.1 parameters: density: 0.69 weight: 0.28 merge_method: ties base_model: nbeerbower/mistral-nemo-bophades-12B parameters: normalize: true int8_mask: true dtype: bfloat16 ``` Stage8c: (Comparable to Stage1 - still needs adjustments but is the best iteration so far despite many other model soups in between) ```yaml models: - model: anthracite-org/magnum-v4-12b parameters: density: 0.82 weight: 0.4 - model: MarinaraSpaghetti/NemoMix-Unleashed-12B parameters: density: 0.69 weight: 0.25 - model: TheDrummer/Rocinante-12B-v1.1 parameters: density: 0.69 weight: 0.35 merge_method: ties base_model: nbeerbower/mistral-nemo-bophades-12B parameters: normalize: true int8_mask: true dtype: bfloat16