LLMs for Accuracy Benchmarking
Precompiled LLMs with --return_logits for Modalix accuracy benchmarking, with SiMa.ai-published results and HF GPU reference baselines.
Text Generation • UpdatedNote Compiled with --return_logits for Modalix accuracy benchmarking.
simaai/LFM2-1.2B-a16w4_return_logits
Text Generation • UpdatedNote Compiled with --return_logits for Modalix accuracy benchmarking.
simaai/LFM2-350M-a16w4_return_logits
Text Generation • UpdatedNote Compiled with --return_logits for Modalix accuracy benchmarking.
simaai/LFM2.5-1.2B-Instruct-a16w4_return_logits
Text Generation • UpdatedNote Compiled with --return_logits for Modalix accuracy benchmarking.
simaai/LFM2.5-1.2B-Thinking-a16w4_return_logits
Text Generation • UpdatedNote Compiled with --return_logits for Modalix accuracy benchmarking.
simaai/LFM2.5-230M-a16w4_return_logits
Text Generation • UpdatedNote Compiled with --return_logits for Modalix accuracy benchmarking.
simaai/LFM2.5-350M-a16w4_return_logits
Text Generation • UpdatedNote Compiled with --return_logits for Modalix accuracy benchmarking.
simaai/Llama-2-7b-chat-hf-a16w4_return_logits
Text Generation • UpdatedNote Compiled with --return_logits for Modalix accuracy benchmarking.
simaai/Llama-3.1-8B-Instruct-a16w4_return_logits
Text Generation • UpdatedNote Compiled with --return_logits for Modalix accuracy benchmarking.
simaai/Llama-3.2-3B-Instruct-a16w4_return_logits
Text Generation • UpdatedNote Compiled with --return_logits for Modalix accuracy benchmarking.
simaai/Mistral-7B-Instruct-v0.3-a16w4_return_logits
Text Generation • UpdatedNote Compiled with --return_logits for Modalix accuracy benchmarking.
simaai/Phi-3.5-mini-instruct-a16w4_return_logits
Text Generation • UpdatedNote Compiled with --return_logits for Modalix accuracy benchmarking.
simaai/gemma-3-1b-it-a16w4_return_logits
Text Generation • UpdatedNote Compiled with --return_logits for Modalix accuracy benchmarking.
simaai/gemma-3-4b-it-a16w4_return_logits
Text Generation • UpdatedNote Compiled with --return_logits for Modalix accuracy benchmarking.