Big (100B+) LLMs (OUTDATED) - a ChuckMcSneed Collection

alpindale/goliath-120b

Text Generation • Updated Dec 21, 2023 • 3.06k • 213

Note Merge of xwin and euryale 1.3. The OG that started the big frankenmerging. Creative and smart.

alpindale/miquella-120b

Text Generation • Updated Feb 1 • 3 • 20

Note Merge of Euryale and Miqu Q5_K_M. Brain damaged(tested only old version).

grimulkan/Goliath-longLORA-120b-rope8-32k-fp16

Text Generation • Updated Jan 27 • 4 • 7

Note 32k context version of Goliath. Dumber and less coherent than the OG Goliath. Has trouble with formatting on short context, has partially lost Alpaca. Confirmed that it works at 17k, but loses minor details of the story, just like the other models extended with this method.

migtissera/Tess-XL-v1.0

Text Generation • Updated Nov 19, 2023 • 2.92k • 13

Note A finetune of Goliath.

ChuckMcSneed/Dicephal-123B

Text Generation • Updated Feb 10 • 2 • 3

Note Self-merge of llama 2 70b. More creative and smarter than the base, but just as disobedient. Should be good for story writing.

ChuckMcSneed/DoubleGold-v0.1-123b-32k

Text Generation • Updated Jan 12 • 3 • 1

Note First 32k context 120b frankenstein. Self-merge of grimulkan/aurelian-alpha0.1-70b-rope8-32K-fp16. Dumber and less coherent than Goliath, but compensates for it with large context. Stays coherent at 17k, not sure how well it works at 32k.

ChuckMcSneed/DoubleGold-v0.5-123b-32k

Text Generation • Updated Jan 27 • 2

Note Self-merge of grimulkan/aurelian-v0.5-70b-rope8-32K-fp16. Like DoubleGold 0.1, but with much less GPTism.

ChuckMcSneed/WinterGoliath-123b

Text Generation • Updated Feb 2 • 778 • 7

Note Merge of Xwin and WinterGoddess. Smarter than Goliath. Likely a bit more aligned.

ChuckMcSneed/WinterGoliath-123b-32k

Text Generation • Updated Jan 30 • 3

Note 32k version of WinterGoliath. Comes with all the problems that the extended context creates. Still smart.

ChuckMcSneed/BenchmaxxxerPS-v1-123b

Text Generation • Updated Feb 3 • 2

Note Meme merge meant to beat the benchmark. Has two personalities: angry hobo and an SJW. Not really worth using, unless you want to talk to a digital schizo.

sophosympatheia/Aurora-Nights-103B-v1.0

Text Generation • Updated Feb 11 • 4 • 11

Note Self-merge of Aurora-Nights-70B-v1.0. Decent model, a bit more "fun" version of Aurora-Nights.

sophosympatheia/Rogue-Rose-103b-v0.2

Text Generation • Updated Feb 11 • 4 • 21

Note This model is a frankenmerge of two custom 70b merges descended from xwin-stellarbright-erp-70b-v2 model.

sophosympatheia/Midnight-Rose-103B-v2.0.3

Text Generation • Updated Feb 17 • 796 • 6

Note Self-merge of sophosympatheia/Midnight-Rose-70B-v2.0.3.

nsfwthrowitaway69/Venus-120b-v1.0

Text Generation • Updated Nov 27, 2023 • 16 • 25

Note Merge of Euryale-1.3-L2-70B, Nous-Hermes-Llama2-70b, and SynthIA-70B-v1.5. Drier than Goliath.

nsfwthrowitaway69/Venus-103b-v1.0

Text Generation • Updated Nov 30, 2023 • 81 • 12

Note A smaller merge of Euryale-1.3-L2-70B, Nous-Hermes-Llama2-70b, and SynthIA-70B-v1.5.

nsfwthrowitaway69/Venus-120b-v1.1

Text Generation • Updated Dec 12, 2023 • 3 • 6

Note A result of interleaving layers of Sao10K/Euryale-1.3-L2-70B, Xwin-LM/Xwin-LM-70B-V0.1, and migtissera/SynthIA-70B-v1.5.

nsfwthrowitaway69/Venus-103b-v1.1

Text Generation • Updated Dec 6, 2023 • 2 • 11

Note A smaller merge of Sao10K/Euryale-1.3-L2-70B, Xwin-LM/Xwin-LM-70B-V0.1, and migtissera/SynthIA-70B-v1.5.

nsfwthrowitaway69/Venus-103b-v1.2

Text Generation • Updated Dec 12, 2023 • 9 • 5

Note A result of interleaving layers of Sao10K/Euryale-1.3-L2-70B and GOAT-AI/GOAT-70B-Storytelling.

nsfwthrowitaway69/Venus-120b-v1.2

Text Generation • Updated Jan 7 • 10 • 49

Note Self-merge of lzlv. Are you going to call all your models "Venus" regardless of what you used to merge? Your naming scheme is leading to confusion.

cognitivecomputations/MegaDolphin-120b

Text Generation • Updated 9 days ago • 4.14k • 66

Note Self-merge of Dolphin-2.2-70b. Very fucking aggresive(GOOD!). Dumber than Goliath.

cognitivecomputations/Samantha-120b

Text Generation • Updated 9 days ago • 7 • 39

Note Self-merge of Samantha-1.11-70b.

Sao10K/Lila-103B-L2

Text Generation • Updated Nov 28, 2023 • 1.43k • 7

Note Self-merge of Lila-70b.

Sao10K/Solus-103B-L2

Text Generation • Updated Nov 28, 2023 • 1.44k • 2

Note Self-merge of Solus-70b.

MarsupialAI/KitchenSink_103b

Text Generation • Updated Mar 16 • 3 • 9

Note Merge of royallab/Aetheria-L2-70B, lizpreciatior/lzlv_70b_fp16_hf, Sao10K/WinterGoddess-1.4x-70B-L2.

ibivibiv/chimera-120b

Text Generation • Updated Jan 4 • 3 • 1

Note This model is a sliced passthrough merge of Sheep Duck, Xwin, and Sythia.

ibivibiv/giant-macaroni-120b

Text Generation • Updated Jan 4 • 4 • 1

Note Merge of AIDC-ai-business/Marcoroni-70B-v1 garage-bAInd/Platypus2-70B-instruct fangloveskari/ORCA_LLaMA_70B_QLoRA.

ibivibiv/athena-120b

Text Generation • Updated Jan 4 • 4 • 1

Note Merge of Sao10K/WinterGoddess-1.4x-70B-L2, Xwin-LM/Xwin-LM-70B-V0.1, and migtissera/SynthIA-70B-v1.5.

llmixer/Xwinter-120b

Text Generation • Updated Jan 17 • 5 • 1

Note A Goliath-120b style frankenmerge of Xwin-LM-70b-v0.1 and WinterGoddess-1.4x-70b. Similar to ChuckMcSneed/WinterGoliath-123b.

llmixer/BigLiz-120b

Text Generation • Updated Mar 12 • 3 • 1

Note A Goliath-120b style frankenmerge of lzlv-70b and WinterGoddess-1.4x-70b.

llmixer/BigAurelian-v0.5-120b-32k

Text Generation • Updated Jan 17 • 4 • 1

Note A Goliath-120b style frankenmerge of aurelian-v0.5-70b-32K and WinterGoddess-1.4x-70b. Not sure if it is really 32k.

llmixer/BigWeave-v15-103b

Text Generation • Updated Mar 5 • 2.64k • 2

Note ???

MatrixC7/Meidebenne-120b-v1.0

Text Generation • Updated Jan 30 • 3

Note This is a self-merge of sophosympatheia/Midnight-Rose-70B-v1.0.

wolfram/miqu-1-120b

Text Generation • Updated Mar 5 • 780 • 47

Note A self-merge of Miqu. Quite good, but has mistralisms. Best 32k model I've tested so far.

wolfram/miquliz-120b

Text Generation • Updated Feb 12 • 2 • 7

Note This is a 120b frankenmerge created by interleaving layers of miqu-1-70b-sf with lzlv_70b_fp16_hf using mergekit.

DiscoResearch/DiscoLM-120b

Text Generation • Updated Dec 13, 2023 • 2.91k • 22

Note A finetune of Goliath.

ludis/tsukasa-120b-qlora

Text Generation • Updated Feb 1 • 3 • 4

Note A finetune of Goliath.

ewof/koishi-120b-qlora-gptq

Text Generation • Updated Dec 17, 2023 • 3 • 1

Note A finetune of Goliath.

dataautogpt3/miqu-120b

Text Generation • Updated Feb 1 • 3 • 5

Note Self-merge of Miqu. Actually 137B.

Dracones/perky-103b-v0.1

Text Generation • Updated Feb 12 • 2 • 1

Note Self-merge of Dracones/perky-70b-v0.1.

abacusai/TheProfessor-155b

Text Generation • Updated Feb 22 • 817 • 90

Note Merge of cognitivecomputations/dolphin-2.2-70b, WizardLM/WizardMath-70B-V1.0, migtissera/SynthIA-70B-v1.2b, epfl-llm/meditron-70b.

deepnight-research/saily_100b

Text Generation • Updated Dec 1, 2023 • 1.46k • 7

Note Frankenmerged model based on Llama2-chat. Actually 118B. Not as bad as Saily 220, but still kinda bad. Gets easily outperformed by some 70Bs.

deepnight-research/Saily_220B

Text Generation • Updated Dec 20, 2023 • 2.95k • 20

Note Actually 208b. A frankenmerge of 10(!) different llama2 finetunes. Not worth using. Has all the problems(fake words, incoherence) of frankenmerges without any of the benefits. Can't even follow the prompt most of the time.