language:
- en
- ko
library_name: transformers
license: cc-by-nc-4.0
pipeline_tag: text-generation
developers: Kanana LLM
training_regime: bf16 mixed precision
base_model: - huihui-ai/kanana-nano-2.1b-instruct-abliterated
tags: - abliterated
- uncensored
Melvin56/kanana-nano-2.1b-instruct-abliterated-GGUF
Original Model : huihui-ai/kanana-nano-2.1b-instruct-abliterated
All quants are made using the imatrix dataset.
| Model | Size (GB) |
|---|---|
| Q2_K_S | 0.914 |
| Q2_K | 0.931 |
| Q3_K_M | 1.138 |
| Q4_K_M | 1.385 |
| Q5_K_M | 1.568 |
| Q6_K | 1.826 |
| Q8_0 | 2.223 |
| F16 | 4.177 |
| F32 | 8.342 |
| CPU (AVX2) | CPU (ARM NEON) | Metal | cuBLAS | rocBLAS | SYCL | CLBlast | Vulkan | Kompute | |
|---|---|---|---|---|---|---|---|---|---|
| K-quants | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ 🐢5 | ✅ 🐢5 | ❌ |
| I-quants | ✅ 🐢4 | ✅ 🐢4 | ✅ 🐢4 | ✅ | ✅ | Partial¹ | ❌ | ❌ | ❌ |
✅: feature works
🚫: feature does not work
❓: unknown, please contribute if you can test it youself
🐢: feature is slow
¹: IQ3_S and IQ1_S, see #5886
²: Only with -ngl 0
³: Inference is 50% slower
⁴: Slower than K-quants of comparable size
⁵: Slower than cuBLAS/rocBLAS on similar cards
⁶: Only q8_0 and iq4_nl