cognitivecomputations MegaDolphin 120b needs about 25.1 GB of memory at Q4_K_M (a 21.9 GB file) with an 8K context window, so it fits on a 32 GB GPU. This repository has 19 quants, from 21.9 GB (Q4_K_M) to 127.8 GB (Q8_0).
cognitivecomputations MegaDolphin 120b is a 120B-parameter Cognitivecomputations model in GGUF format, quantized by RichardErkhov. The repository has 19 quant files, from 21.9 GB (Q4_K_M) to 127.8 GB (Q8_0).
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 21.9 GB | 25.1 GB | Too big | Get file → | |
| Q5_K_M | 24.1 GB | 27.4 GB | Too big | Get file → | |
| Q4_1 | 24.9 GB | 28.4 GB | Too big | Get file → | |
| IQ3_XS | 28.5 GB | 32.4 GB | Too big | Get file → | |
| Q4_0 | 34.4 GB | 39.0 GB | Too big | Get file → | |
| IQ4_XS | 37.7 GB | 42.7 GB | Too big | Get file → | |
| IQ4_NL | 39.4 GB | 44.6 GB | Too big | Get file → | |
| Q2_K | 44.2 GB (2 parts) | 50.0 GB | Too big | Get file → | |
| Q3_K_S | 51.8 GB (2 parts) | 58.5 GB | Too big | Get file → | |
| IQ3_S | 52.0 GB (2 parts) | 58.7 GB | Too big | Get file → | |
| IQ3_M | 53.7 GB (2 parts) | 60.7 GB | Too big | Get file → | |
| Q3_K | 57.8 GB (2 parts) | 65.3 GB | Too big | Get file → | |
| Q3_K_M | 57.8 GB (2 parts) | 65.3 GB | Too big | Get file → | |
| Q3_K_L | 63.0 GB (2 parts) | 71.1 GB | Too big | Get file → | |
| Q5_1 | 67.6 GB (2 parts) | 76.2 GB | Too big | Get file → | |
| Q4_K_S | 68.3 GB (2 parts) | 76.9 GB | Too big | Get file → | |
| Q5_0 | 75.8 GB (2 parts) | 85.4 GB | Too big | Get file → | |
| Q6_K | 98.7 GB (3 parts) | 111.0 GB | Too big | Get file → | |
| Q8_0 | 127.8 GB (3 parts) | 143.7 GB | Too big | Get file → |
What GPU do you need for cognitivecomputations MegaDolphin 120b?
These figures add the file size, an fp16 KV cache at an 8K context window and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: no quant fits at an 8K context.
- 12 GB VRAM: no quant fits at an 8K context.
- 16 GB VRAM: no quant fits at an 8K context.
- 24 GB VRAM: no quant fits at an 8K context.
- 32 GB VRAM: the best quant that fits is Q4_1 (24.9 GB file, about 28.4 GB in use).
- 48 GB VRAM: the best quant that fits is IQ4_NL (39.4 GB file, about 44.6 GB in use).
Which cognitivecomputations MegaDolphin 120b quant should you choose?
Q4_K_M is the usual starting point for cognitivecomputations MegaDolphin 120b: it balances file size and quality well. Here it is a 21.9 GB file.
The Q8_0 file is 127.8 GB, about 5.8 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Frequently asked questions about cognitivecomputations MegaDolphin 120b
How much VRAM does cognitivecomputations MegaDolphin 120b need?
About 25.1 GB at Q4_K_M with an 8K context window. That is the 21.9 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run cognitivecomputations MegaDolphin 120b on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 25.1 GB, so you would need a larger GPU or CPU offloading.
What is the file size of cognitivecomputations MegaDolphin 120b GGUF?
It depends on the quant: Q4_K_M is 21.9 GB, Q5_K_M is 24.1 GB, Q4_1 is 24.9 GB, IQ3_XS is 28.5 GB, Q4_0 is 34.4 GB, IQ4_XS is 37.7 GB and 13 more.
Where can I download cognitivecomputations MegaDolphin 120b GGUF?
The files are on Hugging Face in the RichardErkhov/cognitivecomputations_-_MegaDolphin-120b-gguf repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.