ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-02-24 07:04:11 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add GitHub data: filename sanitization (#640)

Browse Source
This commit is contained in:
Thomas
2025-07-23 13:31:53 +02:00
committed by GitHub
parent 3600d82e98
commit eaa2510a28
626 changed files with 0 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
github-data/pull_requests/555 - Add Falcon-Edge support.md Normal file
View File

@@ -0,0 +1,28 @@
### 🔀 [#555](https://github.com/ikawrakow/ik_llama.cpp/pull/555) - Add Falcon-Edge support
| **Author** | `ikawrakow` |
| :--- | :--- |
| **State** | ❌ **Closed** |
| **Created** | 2025-06-25 |
| **Updated** | 2025-06-26 |
---
#### Description
Closes #551
How to use:
1. Grab a GGUF containing Microsoft's `i2_s` quant packing. E.g.,
```
huggingface-cli download --local-dir falcon tiiuae/Falcon-E-3B-Instruct-GGUF
```
2. Convert to `ik_llama.cpp` quants `iq2_bn` or `iq1_bn`. `iq2_bn` uses 2 bits per weight (bpw), `iq1_bn` uses 1.625 bpw. `iq2_bn` is faster for prompt processing, and may also be faster for token generation (TG) on devices with limited computing power. `iq1_bn` uses 20% less RAM, so that if TG is memory bound, it will be slightly faster than `iq2_bn`. Command to convert is
```
./bin/llama-quantize --allow-requantize falcon/ggml-model-i2_s.gguf falcon_iq2_bn.gguf iq2_bn
```
(replace `iq2_bn` with `iq1_bn` if you prefer the smaller variant.
3. Utilize the just created model file in the usual way with `llama-cli, llama-server`, etc.