Downloads · 30 days
112
9% of all-time downloads
TURKCELL_-_Turkcell-LLM-7b-v1-gguf
RichardErkhov/TURKCELL_-_Turkcell-LLM-7b-v1-gguf
TURKCELL_-_Turkcell-LLM-7b-v1-gguf is a machine learning model from RichardErkhov. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
All-time downloads
1.3K
Public
Repo size
84 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.gguf97.3 GB · 100%
At a glance
- Access
- Public
- Created
- Sep 17, 2024
- Updated
- Sep 17, 2024
- SHA
- 2f678429
Model card
From the Hugging Face model README
Quantization made by Richard Erkhov.
Turkcell-LLM-7b-v1 - GGUF
- Model creator: https://huggingface.co/TURKCELL/
- Original model: https://huggingface.co/TURKCELL/Turkcell-LLM-7b-v1/
| Name | Quant method | Size |
|---|---|---|
| Turkcell-LLM-7b-v1.Q2_K.gguf | Q2_K | 2.6GB |
| Turkcell-LLM-7b-v1.IQ3_XS.gguf | IQ3_XS | 2.89GB |
| Turkcell-LLM-7b-v1.IQ3_S.gguf | IQ3_S | 3.04GB |
| Turkcell-LLM-7b-v1.Q3_K_S.gguf | Q3_K_S | 3.03GB |
| Turkcell-LLM-7b-v1.IQ3_M.gguf | IQ3_M | 3.14GB |
| Turkcell-LLM-7b-v1.Q3_K.gguf | Q3_K | 3.36GB |
| Turkcell-LLM-7b-v1.Q3_K_M.gguf | Q3_K_M | 3.36GB |
| Turkcell-LLM-7b-v1.Q3_K_L.gguf | Q3_K_L | 3.64GB |
| Turkcell-LLM-7b-v1.IQ4_XS.gguf | IQ4_XS | 3.76GB |
| Turkcell-LLM-7b-v1.Q4_0.gguf | Q4_0 | 3.91GB |
| Turkcell-LLM-7b-v1.IQ4_NL.gguf | IQ4_NL | 3.96GB |
| Turkcell-LLM-7b-v1.Q4_K_S.gguf | Q4_K_S | 3.94GB |
| Turkcell-LLM-7b-v1.Q4_K.gguf | Q4_K | 4.16GB |
| Turkcell-LLM-7b-v1.Q4_K_M.gguf | Q4_K_M | 4.16GB |
| Turkcell-LLM-7b-v1.Q4_1.gguf | Q4_1 | 4.33GB |
| Turkcell-LLM-7b-v1.Q5_0.gguf | Q5_0 | 4.75GB |
| Turkcell-LLM-7b-v1.Q5_K_S.gguf | Q5_K_S | 4.75GB |
| Turkcell-LLM-7b-v1.Q5_K.gguf | Q5_K | 4.87GB |
| Turkcell-LLM-7b-v1.Q5_K_M.gguf | Q5_K_M | 4.87GB |
| Turkcell-LLM-7b-v1.Q5_1.gguf | Q5_1 | 5.17GB |
| Turkcell-LLM-7b-v1.Q6_K.gguf | Q6_K | 5.64GB |
| Turkcell-LLM-7b-v1.Q8_0.gguf | Q8_0 | 7.3GB |
Original model description:
license: apache-2.0 language:
- tr
<img src="https://huggingface.co/TURKCELL/Turkcell-LLM-7b-v1/resolve/main/icon.jpeg" alt="Turkcell LLM" width="300"/>
Turkcell-LLM-7b-v1
This model is an extended version of a Mistral-based Large Language Model (LLM) for Turkish. It was trained on a cleaned Turkish raw dataset containing 5 billion tokens. The training process involved using the DORA method initially. Following this, we utilized Turkish instruction sets created from various open-source and internal resources for fine-tuning with the LORA method.
Model Details
- Base Model: Mistral 7B based LLM
- Tokenizer Extension: Specifically extended for Turkish
- Training Dataset: Cleaned Turkish raw data with 5 billion tokens, custom Turkish instruction sets
- Training Method: Initially with DORA, followed by fine-tuning with LORA
DORA Configuration
lora_alpha: 128lora_dropout: 0.05r: 64target_modules: "all-linear"
LORA Fine-Tuning Configuration
lora_alpha: 128lora_dropout: 0.05r: 256target_modules: "all-linear"
Usage Examples
from transformers import AutoModelForCausalLM, AutoTokenizer
device = "cuda" # the device to load the model onto
model = AutoModelForCausalLM.from_pretrained("TURKCELL/Turkcell-LLM-7b-v1")
tokenizer = AutoTokenizer.from_pretrained("TURKCELL/Turkcell-LLM-7b-v1")
messages = [
{"role": "user", "content": "Türkiye'nin başkenti neresidir?"},
]
encodeds = tokenizer.apply_chat_template(messages, return_tensors="pt")
eos_token = tokenizer("<|im_end|>",add_special_tokens=False)["input_ids"][0]
model_inputs = encodeds.to(device)
model.to(device)
generated_ids = model.generate(model_inputs,
max_new_tokens=1024,
do_sample=True,
eos_token_id=eos_token)
decoded = tokenizer.batch_decode(generated_ids)
print(decoded[0])
- Created
- Sep 17, 2024
- Updated
- Sep 17, 2024