Ertas Train base models - a ErtasAI Collection

ErtasAI 's Collections

updated 3 days ago

Open base models mirrored for training in Ertas Studio, in full precision (bf16) and 4-bit. Each card names the original creator and licence.

Upvote

ErtasAI/Qwen3.5-4B

Text Generation • 5B • Updated Jul 19 • 11 • 1

Note Full-precision (bf16) mirror of Qwen/Qwen3.5-4B.
ErtasAI/Qwen3-8B

Text Generation • 8B • Updated Jul 19 • 15

Note Full-precision (bf16) mirror of Qwen/Qwen3-8B.
ErtasAI/Qwen3-8B-bnb-4bit

Text Generation • 8B • Updated Jul 19 • 15

Note 4-bit (bnb-4bit) build of Qwen/Qwen3-8B, for memory-efficient training.
ErtasAI/Qwen3-4B

Text Generation • 4B • Updated Jul 19 • 9

Note Full-precision (bf16) mirror of Qwen/Qwen3-4B.
ErtasAI/Qwen3-4B-bnb-4bit

Text Generation • 4B • Updated Jul 19 • 15

Note 4-bit (bnb-4bit) build of Qwen/Qwen3-4B, for memory-efficient training.
ErtasAI/Qwen2.5-7B-Instruct

Text Generation • 8B • Updated Jul 19 • 6

Note Full-precision (bf16) mirror of Qwen/Qwen2.5-7B-Instruct.
ErtasAI/Qwen2.5-7B-Instruct-bnb-4bit

Text Generation • 8B • Updated Jul 19 • 8 • 1

Note 4-bit (bnb-4bit) build of Qwen/Qwen2.5-7B-Instruct, for memory-efficient training.
ErtasAI/Qwen2.5-3B-Instruct

Text Generation • 3B • Updated Jul 19 • 5

Note Full-precision (bf16) mirror of Qwen/Qwen2.5-3B-Instruct.
ErtasAI/Qwen2.5-3B-Instruct-bnb-4bit

Text Generation • 3B • Updated Jul 19 • 25

Note 4-bit (bnb-4bit) build of Qwen/Qwen2.5-3B-Instruct, for memory-efficient training.
ErtasAI/Qwen2.5-1.5B-Instruct

Text Generation • 2B • Updated Jul 19 • 7

Note Full-precision (bf16) mirror of Qwen/Qwen2.5-1.5B-Instruct.
ErtasAI/Qwen2.5-1.5B-Instruct-bnb-4bit

Text Generation • 2B • Updated Jul 19 • 8

Note 4-bit (bnb-4bit) build of Qwen/Qwen2.5-1.5B-Instruct, for memory-efficient training.
ErtasAI/Qwen2.5-0.5B-Instruct

Text Generation • 0.5B • Updated Jul 19 • 9

Note Full-precision (bf16) mirror of Qwen/Qwen2.5-0.5B-Instruct.
ErtasAI/Qwen2.5-0.5B-Instruct-bnb-4bit

Text Generation • 0.5B • Updated Jul 19 • 13 • 1

Note 4-bit (bnb-4bit) build of Qwen/Qwen2.5-0.5B-Instruct, for memory-efficient training.
ErtasAI/gemma-4-E2B-it

Text Generation • 5B • Updated Jul 19 • 4

Note Full-precision (bf16) mirror of google/gemma-4-E2B-it.
ErtasAI/gemma-4-E2B-it-bnb-4bit

Text Generation • 5B • Updated Jul 19 • 15

Note 4-bit (bnb-4bit) build of google/gemma-4-E2B-it, for memory-efficient training.
ErtasAI/gemma-3-4b-it

Text Generation • 4B • Updated Jul 19 • 7

Note Full-precision (bf16) mirror of google/gemma-3-4b-it.
ErtasAI/gemma-3-4b-it-bnb-4bit

Text Generation • 4B • Updated Jul 19 • 6

Note 4-bit (bnb-4bit) build of google/gemma-3-4b-it, for memory-efficient training.
ErtasAI/gemma-3-1b-it

Text Generation • 1.0B • Updated Jul 19 • 8

Note Full-precision (bf16) mirror of google/gemma-3-1b-it.
ErtasAI/gemma-3-1b-it-bnb-4bit

Text Generation • 1B • Updated Jul 19 • 7

Note 4-bit (bnb-4bit) build of google/gemma-3-1b-it, for memory-efficient training.
ErtasAI/Meta-Llama-3.1-8B-Instruct

Text Generation • 8B • Updated Jul 19 • 5

Note Full-precision (bf16) mirror of meta-llama/Llama-3.1-8B-Instruct.
ErtasAI/Meta-Llama-3.1-8B-Instruct-bnb-4bit

Text Generation • 8B • Updated Jul 19 • 14

Note 4-bit (bnb-4bit) build of meta-llama/Llama-3.1-8B-Instruct, for memory-efficient training.
ErtasAI/Llama-3.2-3B-Instruct

Text Generation • 3B • Updated Jul 19 • 9

Note Full-precision (bf16) mirror of meta-llama/Llama-3.2-3B-Instruct.
ErtasAI/Llama-3.2-3B-Instruct-bnb-4bit

Text Generation • 3B • Updated Jul 19 • 21

Note 4-bit (bnb-4bit) build of meta-llama/Llama-3.2-3B-Instruct, for memory-efficient training.
ErtasAI/Llama-3.2-1B-Instruct

Text Generation • 1B • Updated Jul 19 • 9

Note Full-precision (bf16) mirror of meta-llama/Llama-3.2-1B-Instruct.
ErtasAI/Llama-3.2-1B-Instruct-bnb-4bit

Text Generation • 1B • Updated Jul 19 • 28

Note 4-bit (bnb-4bit) build of meta-llama/Llama-3.2-1B-Instruct, for memory-efficient training.
ErtasAI/LFM2.5-1.2B-Instruct

Text Generation • 1B • Updated Jul 19 • 16 • 2

Note Full-precision (bf16) mirror of LiquidAI/LFM2.5-1.2B-Instruct.
ErtasAI/LFM2.5-350M

Text Generation • 0.4B • Updated Jul 19 • 7

Note Full-precision (bf16) mirror of LiquidAI/LFM2.5-350M.
ErtasAI/LFM2.5-230M

Text Generation • 0.2B • Updated Jul 19 • 15

Note Full-precision (bf16) mirror of LiquidAI/LFM2.5-230M.
ErtasAI/Phi-4-mini-instruct

Text Generation • 4B • Updated Jul 19 • 7

Note Full-precision (bf16) mirror of microsoft/Phi-4-mini-instruct.
ErtasAI/Phi-4-mini-instruct-bnb-4bit

Text Generation • 4B • Updated Jul 19 • 28 • 1

Note 4-bit (bnb-4bit) build of microsoft/Phi-4-mini-instruct, for memory-efficient training.
ErtasAI/Phi-3.5-mini-instruct

Text Generation • 4B • Updated Jul 19 • 6

Note Full-precision (bf16) mirror of microsoft/Phi-3.5-mini-instruct.
ErtasAI/Phi-3.5-mini-instruct-bnb-4bit

Text Generation • 4B • Updated Jul 19 • 9 • 1

Note 4-bit (bnb-4bit) build of microsoft/Phi-3.5-mini-instruct, for memory-efficient training.
ErtasAI/mistral-7b-instruct-v0.3

Text Generation • 7B • Updated Jul 19 • 7

Note Full-precision (bf16) mirror of mistralai/Mistral-7B-Instruct-v0.3.
ErtasAI/mistral-7b-instruct-v0.3-bnb-4bit

Text Generation • 7B • Updated Jul 19 • 7

Note 4-bit (bnb-4bit) build of mistralai/Mistral-7B-Instruct-v0.3, for memory-efficient training.
ErtasAI/DeepSeek-R1-Distill-Qwen-7B

Text Generation • 8B • Updated Jul 19 • 9

Note Full-precision (bf16) mirror of deepseek-ai/DeepSeek-R1-Distill-Qwen-7B.
ErtasAI/DeepSeek-R1-Distill-Qwen-7B-bnb-4bit

Text Generation • 8B • Updated Jul 19 • 11

Note 4-bit (bnb-4bit) build of deepseek-ai/DeepSeek-R1-Distill-Qwen-7B, for memory-efficient training.

Upvote