RichardErkhov/

Ba2han_-_Pretraining_Turkish-test-5500s-exl2

Quantization made by Richard Erkhov.

Pretraining_Turkish-test-5500s - EXL2

| Branch | Bits | Description | | ----- | ---- | ------- | ------ | ------ | ------ | ------ | ------------ | | 8_0 | 8.0 | Maximum quality that ExLlamaV2 can produce, near unquantized performance. | | 6_5 | 6.5 | Very similar to 8.0, good tradeoff of size vs performance, recommended. | | 5_0 | 5.0 | Slightly lower quality vs 6.5, but usable on 8GB cards. | | 4_25 | 4.25 | GPTQ equivalent bits per weight, slightly higher quality. | | 3_5 | 3.5 | Lower quality, only use if you have to. |

With git:

With huggingface hub:

To download a specific branch, use the --revision parameter. For example, to download the 6.5 bpw branch: Linux:

Windows (which apparently doesn't like _ in folders sometimes?):

license: mit language:

Pretrained on 1B (mostly Turkish) tokens from HF and "high quality" scraped data using 1 RTX 3090. The training will continue. The model already can produce sensible sentences in Turkish.

HF kaynaklı ve scrape edilen 1 Milyar (çoğunlukla Türkçe) token ile 1 RTX 3090 kullanılarak eğitilmiştir. Eğitim devam edecek. Model şimdiden düzgün Türkçe cümleler kurabiliyor:

(top_k=24, repetition_penalty=1.1, temperature=0.12, seed=1022)