Llama 2 70B HF vs Selfrag Llama2 7B
Side-by-side comparison of VRAM requirements, quantization, context length, and hardware compatibility.
Specifications
| Llama 2 70B HF | Selfrag Llama2 7B | |
|---|---|---|
| Parameters | 69.0B | 7B |
| Context | — | 4K |
| Architecture | — | LlamaForCausalLM |
| License | Llama 2 Community | MIT |
| Downloads | 15.4K | 2.8K |
| Released | Apr 2024 | Jan 2024 |
VRAM by Quantization: Llama 2 70B HF vs Selfrag Llama2 7B
| Quantization | Bits | Llama 2 70B HF VRAM | Selfrag Llama2 7B VRAM |
|---|---|---|---|
| BF16 | 16.00 | 151.8 GB | 15.4 GB |
Verdict
Selfrag Llama2 7B needs less VRAM at BF16 (15.4 GB vs 151.8 GB), so it fits on smaller GPUs. Llama 2 70B HF is the more widely downloaded of the two.
Frequently Asked Questions
- Which needs less VRAM, Llama 2 70B HF or Selfrag Llama2 7B?
At BF16, Llama 2 70B HF needs 151.8 GB and Selfrag Llama2 7B needs 15.4 GB, so Selfrag Llama2 7B is the lighter option to run locally.
- What is the difference between Llama 2 70B HF and Selfrag Llama2 7B?
Llama 2 70B HF is a 69.0B model from Meta (Llama 2 family), while Selfrag Llama2 7B is a 7B model from selfrag (Llama 2 family). Compare their VRAM requirements above to see which fits your GPU or Mac.