Update README.md
Browse files
README.md
CHANGED
|
@@ -17,22 +17,22 @@ pipeline_tag: text-generation
|
|
| 17 |
|
| 18 |
# Kumru-2B LoRA Adapter
|
| 19 |
|
| 20 |
-
|
| 21 |
-
`vngrs-ai/Kumru-2B
|
| 22 |
-
|
| 23 |
-
|
| 24 |
|
| 25 |
-
## Model
|
| 26 |
|
| 27 |
-
- **
|
| 28 |
-
- **
|
| 29 |
-
- **
|
| 30 |
-
- **LoRA rank / alpha:**
|
| 31 |
-
- **
|
| 32 |
-
-
|
| 33 |
-
- **
|
| 34 |
|
| 35 |
-
##
|
| 36 |
|
| 37 |
```python
|
| 38 |
from peft import PeftModel
|
|
@@ -53,22 +53,23 @@ pipeline_tag: text-generation
|
|
| 53 |
outputs = model.generate(inputs, max_new_tokens=512, temperature=0.7, top_p=0.9)
|
| 54 |
print(tokenizer.decode(outputs[0][inputs.shape[-1]:], skip_special_tokens=True))
|
| 55 |
```
|
| 56 |
-
> Not:
|
| 57 |
|
| 58 |
-
##
|
| 59 |
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
|
|
|
| 63 |
|
| 64 |
- Script: export_kumru.py
|
| 65 |
|
| 66 |
-
##
|
| 67 |
|
| 68 |
-
- Kumru-2B
|
| 69 |
-
-
|
| 70 |
-
-
|
| 71 |
-
|
| 72 |
### Framework versions
|
| 73 |
|
| 74 |
- PEFT 0.11.1
|
|
|
|
| 17 |
|
| 18 |
# Kumru-2B LoRA Adapter
|
| 19 |
|
| 20 |
+
This repository provides a **LoRA** adapter distilled from the **VNGRS Kumru-2B** model (
|
| 21 |
+
`vngrs-ai/Kumru-2B`, the SFT/chat variant) to be applied on top of the base model
|
| 22 |
+
`vngrs-ai/Kumru-2B-Base`. The goal is to transfer Kumru’s chat/instruction behavior
|
| 23 |
+
to `Kumru-2B-Base` deployments with a lightweight file footprint.
|
| 24 |
|
| 25 |
+
## Model Summary
|
| 26 |
|
| 27 |
+
- **Base model:** `vngrs-ai/Kumru-2B-Base`
|
| 28 |
+
- **Source (target behavior) model:** `vngrs-ai/Kumru-2B` (SFT/chat)
|
| 29 |
+
- **echnique:** Low-Rank Adaptation (LoRA)
|
| 30 |
+
- **LoRA rank / alpha:** 768 / 1024 _(update these if you produce a different build)_
|
| 31 |
+
- **Layer coverage:** All self-attention and MLP projections
|
| 32 |
+
- **Output artifacts:** PEFT-compatible `adapter_config.json` + `adapter_model.safetensors`
|
| 33 |
+
- **License:** Apache 2.0 (aligned with VNGRS Kumru model licensing)
|
| 34 |
|
| 35 |
+
## Usage
|
| 36 |
|
| 37 |
```python
|
| 38 |
from peft import PeftModel
|
|
|
|
| 53 |
outputs = model.generate(inputs, max_new_tokens=512, temperature=0.7, top_p=0.9)
|
| 54 |
print(tokenizer.decode(outputs[0][inputs.shape[-1]:], skip_special_tokens=True))
|
| 55 |
```
|
| 56 |
+
> Not: This adapter must be used together with `vngrs-ai/Kumru-2B-Base`.
|
| 57 |
|
| 58 |
+
## Extraction Process
|
| 59 |
|
| 60 |
+
The adapter is obtained by computing the delta between the base and the SFT checkpoints and factorizing it with **SVD**
|
| 61 |
+
into low-rank components. In this release, the measured reconstruction error is approximately **0.409**. To better preserve
|
| 62 |
+
quality, you may increase rank/alpha and export a new version (e.g., rank **1024** / alpha **2048**). A lower-error build will
|
| 63 |
+
be added as soon as possible.
|
| 64 |
|
| 65 |
- Script: export_kumru.py
|
| 66 |
|
| 67 |
+
## Known Limitations
|
| 68 |
|
| 69 |
+
- Kumru-2B is still a ~2B-parameter model; it may struggle with very long context, rare technical terms, and complex math.
|
| 70 |
+
- With low ranks, SVD-based LoRA can be less stable than the original SFT checkpoint.
|
| 71 |
+
- Training data is based on VNGRS’s public Turkish corpus cleaning pipeline; truthfulness/hallucination issues may still occur.
|
| 72 |
+
|
| 73 |
### Framework versions
|
| 74 |
|
| 75 |
- PEFT 0.11.1
|