Võrdle
Llama Guard 3 8B vs Llama Guard 4 12B
Aseta kaks kuni neli mudelit kõrvuti väljadel, mis määravad kasutuselevõtu: kontekst, hind, piirkond, litsents, keeled, majutamine ja see, mis on kontrollitud.
| Llama Guard 3 8B | Llama Guard 4 12B | |
|---|---|---|
| Pakkuja | Meta | Meta |
| Parameetrid | 8B | 12B |
| Kontekstaken | Kinnitamata: pakkuja ei ole seda väidet dokumenteerinud. | Kinnitamata: pakkuja ei ole seda väidet dokumenteerinud. |
| Litsents | llama3.1 | other |
| Modaalsused | text | text, vision |
| Väljalaske kuupäev | 22. juuli 2024 | 23. apr 2025 |
| Keeled | Tugev: en Piisav: — | Tugev: en Piisav: — |
| Avatud kaalud | jah | jah |
| Ülesanded | Tundlikud andmed, Klienditugi | Tundlikud andmed, Nägemine, Klienditugi |
Kui kaks mudelit erinevad, näitab see tabel erinevust ja lõpetab. LLM EU ei avalda võitjamärki ilma selle ülesande dokumenteeritud skoorita ning puuduv näitaja jääb puuduma, mitte ei muutu nulliks.
Millal kasutada
Llama Guard 3 8B
Tugevused
- open weights are published on Hugging Face and can be downloaded without a vendor agreement
- at 8B parameters it fits on a single accelerator
- the model page lists en
- instruction-tuned checkpoint with a chat template in the repository
Piirangud
- catalogue only — LLM EU does not run this model, and nothing on this card implies an LLM EU endpoint
- no evaluation has been run by LLM EU
- the published configuration records no context window, so the card carries none
- at 8B parameters, the weights fit on a single modest accelerator
- a classifier, not a chat model: it labels prompts and responses against a safety taxonomy instead of answering them
Ülesanded
Llama Guard 4 12B
Tugevused
- open weights are published on Hugging Face and can be downloaded without a vendor agreement
- at 12B parameters it fits on a single accelerator
- the model page lists en
- instruction-tuned checkpoint with a chat template in the repository
Piirangud
- catalogue only — LLM EU does not run this model, and nothing on this card implies an LLM EU endpoint
- no evaluation has been run by LLM EU
- the llama4 (other) licence is a custom agreement whose commercial terms were not established from its published text, so commercial serving is not permitted until a human clears it; the router must not route to it
- the published configuration records no context window, so the card carries none
- at 12B parameters, self-hosting needs a single accelerator with enough memory for the weights
- vision checkpoints are heavier to serve than their parameter count alone suggests, because the image encoder and pre-processing run alongside the language model
- a classifier, not a chat model: it labels prompts and responses against a safety taxonomy instead of answering them
Ülesanded
Ülesanded: Tundlikud andmed, Klienditugi