Posts tagged small language models collect Will Spurlock's writing on this topic. The three newest excerpts: Microsoft's Phi-4 14B delivers GPT-4o-level reasoning and beats Llama 3.3 70B on math and coding benchmarks—proving data quality beats raw parameter count. Mistral launches Ministral 3B and 8B—best-in-class edge models with 128k context, outperforming Llama 3.2 3B and Llama 3.1 8B at breakthrough pricing. Phi-3.5-mini, MoE, and vision ship under MIT with 128K context—Microsoft’s SLMs benchmark beside Llama-3.1-8B on multilingual and long-context suites.
3 posts as of 2024-12-12, counted from content/blog frontmatter
Discover insights and strategies for leveraging technology in business
Posts tagged small language models collect Will Spurlock's writing on this topic. The three newest excerpts: Microsoft's Phi-4 14B delivers GPT-4o-level reasoning and beats Llama 3.3 70B on math and coding benchmarks—proving data quality beats raw parameter count. Mistral launches Ministral 3B and 8B—best-in-class edge models with 128k context, outperforming Llama 3.2 3B and Llama 3.1 8B at breakthrough pricing. Phi-3.5-mini, MoE, and vision ship under MIT with 128K context—Microsoft’s SLMs benchmark beside Llama-3.1-8B on multilingual and long-context suites.
3 posts as of 2024-12-12, counted from content/blog frontmatter.
Start with "Microsoft Phi-4 14B: Beating Llama 3.3 70B at 5× Fewer Parameters". Microsoft's Phi-4 14B delivers GPT-4o-level reasoning and beats Llama 3.3 70B on math and coding benchmarks—proving data quality beats raw parameter count.
