Climate-domain continued pretraining
According to the model card, the 70B decoder continued from Llama-2 on 4.2B tokens of curated climate documents, then instruction-tuned on expert-collected pairs.
Open Source Model Profile · eci-io
climategpt-70b is a 69B-parameter Llama climate-science fine-tune from eci-io. Its model card documents continued pretraining on 4.2B climate tokens with expert instruction tuning.
climategpt-70b is published by eci-io as a Llama-family text-generation model. The captured configuration identifies LlamaForCausalLM, and Safetensors metadata reports 68,978,745,344 parameters (about 69B). According to the model card, it adapts Llama-2-70B to climate science with 4.2B tokens of curated climate documents plus expert instruction tuning.
According to the model card, the 70B decoder continued from Llama-2 on 4.2B tokens of curated climate documents, then instruction-tuned on expert-collected pairs.
The card describes a ChatML-formatted English question-answering model trained for retrieval augmentation with up to 5 references in context.
According to the model card, training used 8 NVIDIA H100 GPUs for 2,182 hours on hydropower, emitting 40.6kg CO2eq.
Source: eci-io/climategpt-70b
Captured: Unknown. Processed: 2026-09-07T19:34:43.972435+00:00.
ClimateGPT-70B ClimateGPT is a family of AI models designed to synthesize interdisciplinary research on climate change. ClimateGPT-70B is a 70 billion parameter transformer decoder model that was adapted from Llama-2 to the domain of climate science using continuous pre-training on a collection of 4.2B tokens from curated climate documents. The model is further instruction fine-tuned on a dataset of instruction-completion pairs manually collected by AppTek in cooperation with climate scientists. ClimateGPT-7B outperforms Llama-2-70B Chat on our climate-specific benchmarks. The model is designed to be used together with retrieval aug…
F001F002F003F004F005F006F007F009F010F011F012F013F015F016F017F018F020F021