Giga ML's X1 Large 32k - The most powerful on-prem deployable model

We are announcing our first lineup of on-premise LLMs, X1 Large 8k, 32k — pre-trained and fine-tuned versions of llama2 70B, which are outperforming Claude 2 on the MT bench with a score of 8.1 vs 8. (White paper coming soon with performance on all the benchmarks.)

X1 Large is available for further fine-tuning and pre-training. Try it out here and let us know what you think!

The problems today:

1. Pre-training: Existing large language models (LLMs) lack the ability to pre-train on specific text data, hindering their effectiveness in specialized domains like healthcare, legal, and finance.
2. Fine-tuning: The inability to fine-tune LLMs for specific output structures or forms restricts their adaptability in critical areas requiring tailored responses.
3. Privacy: Organizations dealing with sensitive customer data face trust & compliance challenges when using third-party servers like OpenAI and Anthropic.

X1 Large:

• Performance: Achieves an MT bench score of 8.1, surpassing Claude 2 after fine-tuning and pre-training.
• Customization: Our unique pre-training and fine-tuning capabilities provide unrivaled performance for industry-specific use cases.
• Security: Offers secure on-premise deployment, ensuring data privacy for enterprises.
• State of the art RAG: We’re partnering with @Mano AI to bring state of art RAG for on-prem deployment on your petabytes of data.

Our ask:

Try out out the demo here with your prompt and let us know how it performed! Email us at founders@gigaml.com if you want fine-tuning and pre-training access in-cloud or on-premise.

Coming soon:

X1 Large Med: Continuing pre training on Medical data.

X1 Large Law: Continuing pre training on entire law data base of all the countries.