Architecture
data/models.yaml. Every label is auditable
against the model's sources.
Specs
- Architecture
- dense
- Total params
- 180B
- Active params
- 180B
- Context window
- 2K tokens
- Attention
- mqa
- Position encoding
- rope
- Training hardware
- A100 40GB
- Post-training
- sft
- OSI-approved
- no
- Data released
- no
- Training code
- not released
Recommended use cases
- large-scale generation under TII license
- research deployments with 8x A100 80GB
Available quantizations
GGUF llama.cpp's container; the common local format, k-quants from Q2 to Q8.
runs on llama.cpp, Ollama
Verified via the Hugging Face model tree ↗. Community quantizations change over time; the families shown are those with published weights at audit time.
Notable innovations
- · Largest openly available weights at 2023 release
- · Multi-query attention with parallel attention/MLP blocks
- · RefinedWeb pretraining corpus
Known limitations
Lineage
Chat fine-tune of Falcon-180B base; descends from the earlier Falcon-7B and Falcon-40B releases.