unsloth
LFM2.5 1.2B Instruct
LFM2.5 1.2B Instruct is a 1.17B dense lfm2 model from unsloth. Configured context length is 128,000 tokens. License: other. Quant sizes are the GGUF file bytes published on Hugging Face when a file exists.
Parameters1.2B
Context125K
Attentiongqa
Licenseother
16 layers · 8 KV heads · head dim 64 · GGUF repo · config
Quant file sizes
| Quant | File size | Source |
|---|---|---|
| BF16 | 2.18 GB | Hugging Face file bytes |
| UD-Q8_K_XL | 1.28 GB | Hugging Face file bytes |
| Q8_0 | 1.16 GB | Hugging Face file bytes |
| UD-Q6_K_XL | 0.93 GB | Hugging Face file bytes |
| Q6_K | 0.90 GB | Hugging Face file bytes |
| UD-Q5_K_XL | 0.79 GB | Hugging Face file bytes |
| Q5_K_M | 0.79 GB | Hugging Face file bytes |
| Q5_K_S | 0.77 GB | Hugging Face file bytes |
| UD-Q4_K_XL | 0.68 GB | Hugging Face file bytes |
| Q4_K_M | 0.68 GB | Hugging Face file bytes |
| Q4_K_S | 0.65 GB | Hugging Face file bytes |
| Q4_1 | 0.71 GB | Hugging Face file bytes |
| Q4_0 | 0.65 GB | Hugging Face file bytes |
| UD-Q3_K_XL | 0.56 GB | Hugging Face file bytes |
| Q3_K_M | 0.56 GB | Hugging Face file bytes |
| Q3_K_S | 0.52 GB | Hugging Face file bytes |
| UD-Q2_K_XL | 0.45 GB | Hugging Face file bytes |
| Q2_K | 0.90 GB | Hugging Face file bytes |
On common machines
| Device | Verdict | Quant | Speed |
|---|---|---|---|
| GeForce RTX 4060 Ti 8GB | Runs great | BF16 | 61–83 |
| GeForce RTX 3060 12GB | Runs great | BF16 | 76–103 |
| GeForce RTX 4060 Ti 16GB | Runs great | BF16 | 61–83 |
| GeForce RTX 4090 | Runs great | BF16 | 212–289 |
| GeForce RTX 5090 | Runs great | BF16 | 377–515 |
| Apple M4 Pro | Runs great | BF16 | 57–78 |
| Apple M5 | Runs great | BF16 | 32–44 |
| CPU only · DDR5-5600 | Runs on CPU | BF16 | 19–26 |
Smallest devices that fit it fully
- GeForce RTX 3050 6GB — Runs great, BF16, 35–48 tok/s
- GeForce RTX 2060 6GB — Runs great, BF16, 71–96 tok/s
- GeForce GTX 1660 Super — Runs great, BF16, 71–96 tok/s
- GeForce GTX 1660 Ti — Runs great, BF16, 61–83 tok/s
- Intel Arc A380 — Runs great, BF16, 39–53 tok/s
- GeForce RTX 5060 Ti 8GB — Runs great, BF16, 94–129 tok/s
Run it
Commands for the GeForce RTX 4060 Ti 16GB estimate (BF16).
llama.cpp
llama-server -hf unsloth/LFM2.5-1.2B-Instruct-GGUF:BF16 -c 8192 -ngl 16
Ollama
ollama run lfm2.5
LM Studio
lms get unsloth/LFM2.5-1.2B-Instruct-GGUF # Then load quant BF16