The Yi series models are the next generation of open-source large language models trained from scratch by 01.AI. Yi-34B model ranked first among all existing open-source models (such as Falcon-180B, Llama-70B, Claude) in both English and Chinese on various benchmarks, including Hugging Face Open LLM Leaderboard (pre-trained) and C-Eval (based on data available up to November 2023).
On-demand DeploymentDocs | On-demand deployments give you dedicated GPUs for Yi 34B using Fireworks' reliable, high-performance system with no rate limits. |
Yi 34B is a 34.4 billion parameter base language model developed by 01.AI. It is part of the Yi series, trained from scratch to support both English and Chinese. As of late 2023, it ranked first among open-source models (including Falcon-180B, Llama 2-70B, and Claude) on benchmarks like the Hugging Face Open LLM Leaderboard and C-Eval.
The model is suitable for:
Yi 34B supports a context length of 4,096 tokens.
The full 4.1K token window is available when running the model on Fireworks' on-demand infrastructure.
Outputs are constrained by the 4.1K token context length (prompt + completion combined).
Yi 34B has 34.4 billion parameters.
Standard fine-tuning is not supported, but LoRA (parameter-efficient fine-tuning) is supported via Fireworks' Serverless LoRA framework.
Token billing is based on total input + output token usage.
