mirror of
https://github.com/ikawrakow/ik_llama.cpp.git
synced 2026-01-30 11:09:51 +00:00
25 lines
1.4 KiB
Markdown
25 lines
1.4 KiB
Markdown
### 🔀 [#106](https://github.com/ikawrakow/ik_llama.cpp/pull/106) - Bitnet changes
|
|
|
|
| **Author** | `ikawrakow` |
|
|
| :--- | :--- |
|
|
| **State** | ❌ **Closed** |
|
|
| **Created** | 2024-10-24 |
|
|
| **Updated** | 2024-10-25 |
|
|
|
|
---
|
|
|
|
#### Description
|
|
|
|
* Change `IQ1_BN` and `IQ2_BN` to have per row scales. In that way we can handle Bitnet models with and without separate tensor scales
|
|
* Remove `IQ1_TN` and `IQ2_TN`. With the above change these are now redundant. `IQ1_BN` and `IQ2_BN` are also faster, so no reason to keep these around
|
|
* Change `build_bitnet()` to use the standard `llm_build_kv()` function for the self attention portion. I was hoping this would also allow to use FA, but nope, the Bitnet models have a strange head size of 100 that is not supported by the FA implementations.
|
|
|
|
Everything works except - can you guess? - Metal. There is something wrong with the dot product kernels and I simply don't see what. I have to fix Metal before merging.
|
|
|
|
On CUDA (RTX-4080) we now get 368 t/s for TG-128 with the 3.3B Bitnet model (`IQ2_BN`). When I first added Bitnet support we were at ~320 t/s, so quite an improvement since then.
|
|
|
|
**Update**
|
|
|
|
I wasted quite some time trying to figure out why the Bitnet changes don't work on Metal. At the end it turned out that it is PR #98 that breaks the Metal back-end. So, this PR reverts #98.
|
|
|
|
@agray3 Do you have the ability to investigate why #98 breaks the Metal back-end? |