GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
dsfsi
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
dsfsi
Model Tree
google/gemma-2-9b-it
Input Modalities
Output Modalities
Supported Functionality
| Property | Value |
|---|---|
| Base Model | google/gemma-2-9b-it |
| Translation Direction | Hausa → English |
| LoRA Rank (r) | 64 |
| LoRA Alpha | 128 |
| Training Method | QLoRA (4-bit quantization) |
| Domain | Scientific/Academic texts |
LoRA (Low-Rank Adaptation) enables efficient fine-tuning by training only a small number of additional parameters. This adapter adds only ~32.0M parameters to the base model while achieving strong translation performance.
Performance on the AfriScience-MT test set:
| Split | BLEU | chrF | SSA-COMET |
|---|---|---|---|
| Test | - | - | - |
Metrics explanation:
python
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfigfrom peft import PeftModelimport torch # Configure 4-bit quantization (recommended for memory efficiency)bnb_config = BitsAndBytesConfig( load_in_4bit=True, bnb_4bit_compute_dtype=torch.bfloat16, bnb_4bit_quant_type="nf4", bnb_4bit_use_double_quant=True,) # Load base modelbase_model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b-it", quantization_config=bnb_config, device_map="auto", torch_dtype=torch.bfloat16,)tokenizer = AutoTokenizer.from_pretrained("google/gemma-2-9b-it") # Load LoRA adapteradapter_name = "dsfsi/gemma_2_9b_it-lora-r64-hau-eng"model = PeftModel.from_pretrained(base_model, adapter_name)model.eval() # Prepare translation promptsource_text = "Climate change significantly impacts agricultural productivity in sub-Saharan Africa."instruction = "Translate the following Hausa scientific text to English." # Format for Gemma chat templatemessages = [{"role": "user", "content": f"{instruction}\n\n{source_text}"}]prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True) # Generate translationinputs = tokenizer(prompt, return_tensors="pt").to(model.device)with torch.no_grad(): outputs = model.generate( **inputs, max_new_tokens=256, num_beams=5, early_stopping=True, pad_token_id=tokenizer.pad_token_id, ) # Decode only the generated partgenerated = outputs[0][inputs["input_ids"].shape[1]:]translation = tokenizer.decode(generated, skip_special_tokens=True)print(translation)python
# For GPUs with sufficient memory (>24GB for larger models)base_model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b-it", device_map="auto", torch_dtype=torch.bfloat16,)model = PeftModel.from_pretrained(base_model, "dsfsi/gemma_2_9b_it-lora-r64-hau-eng")| Parameter | Value |
|---|---|
| LoRA Rank (r) | 64 |
| LoRA Alpha | 128 |
| LoRA Dropout | 0.05 |
| Target Modules | q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj |
| Epochs | 3 |
| Batch Size | 2 |
| Learning Rate | 2e-04 |
| Max Sequence Length | 512 |
| Gradient Accumulation | 4 |
| Configuration | VRAM Required |
|---|---|
| 4-bit (QLoRA) | ~8-12 GB |
| 8-bit | ~16-20 GB |
| Full precision | ~24-40 GB |
To reproduce this adapter:
bash
# Clone the AfriScience-MT repositorygit clone https://github.com/afriscience-mt/afriscience-mt.gitcd afriscience-mt # Install dependenciespip install -r requirements.txt # Run LoRA trainingpython -m afriscience_mt.scripts.run_lora_training \ --data_dir ./data \ --source_lang hau \ --target_lang eng \ --model_name google/gemma-2-9b-it \ --model_type gemma \ --lora_rank 64 \ --output_dir ./output \ --num_epochs 3 \ --batch_size 4 \ --load_in_4bitIf you use this model, please cite the AfriScience-MT paper (arXiv:2605.29741):
bibtex
@article{abdulmumin2026afriscience, title = {AfriScience-MT: Towards Decolonizing Science in Africa through Text Translation}, author = {Abdulmumin, Idris and Gwadabe, Tajuddeen and Muhammad, Shamsuddeen Hassan and Adelani, David Ifeoluwa and Khalo, Nomonde and Ahmad, Ibrahim Said and Modupe, Abiodun and Mumm, Anina and Biyela, Sibusiso and Rabie, Michelle and Havemann, Johanna and Rei, Marek and Abbott, Jade and Marivate, Vukosi}, journal = {arXiv preprint arXiv:2605.29741}, year = {2026}, url = {https://arxiv.org/abs/2605.29741}}This adapter is released under the Apache 2.0 License.
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfigfrom peft import PeftModelimport torch # Configure 4-bit quantization (recommended for memory efficiency)bnb_config = BitsAndBytesConfig( load_in_4bit=True, bnb_4bit_compute_dtype=torch.bfloat16, bnb_4bit_quant_type="nf4", bnb_4bit_use_double_quant=True,) # Load base modelbase_model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b-it", quantization_config=bnb_config, device_map="auto", torch_dtype=torch.bfloat16,)tokenizer = AutoTokenizer.from_pretrained("google/gemma-2-9b-it") # Load LoRA adapteradapter_name = "dsfsi/gemma_2_9b_it-lora-r64-hau-eng"model = PeftModel.from_pretrained(base_model, adapter_name)model.eval() # Prepare translation promptsource_text = "Climate change significantly impacts agricultural productivity in sub-Saharan Africa."instruction = "Translate the following Hausa scientific text to English." # Format for Gemma chat templatemessages = [{"role": "user", "content": f"{instruction}\n\n{source_text}"}]prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True) # Generate translationinputs = tokenizer(prompt, return_tensors="pt").to(model.device)with torch.no_grad(): outputs = model.generate( **inputs, max_new_tokens=256, num_beams=5, early_stopping=True, pad_token_id=tokenizer.pad_token_id, ) # Decode only the generated partgenerated = outputs[0][inputs["input_ids"].shape[1]:]translation = tokenizer.decode(generated, skip_special_tokens=True)print(translation)# For GPUs with sufficient memory (>24GB for larger models)base_model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b-it", device_map="auto", torch_dtype=torch.bfloat16,)model = PeftModel.from_pretrained(base_model, "dsfsi/gemma_2_9b_it-lora-r64-hau-eng")# Clone the AfriScience-MT repositorygit clone https://github.com/afriscience-mt/afriscience-mt.gitcd afriscience-mt # Install dependenciespip install -r requirements.txt # Run LoRA trainingpython -m afriscience_mt.scripts.run_lora_training \ --data_dir ./data \ --source_lang hau \ --target_lang eng \ --model_name google/gemma-2-9b-it \ --model_type gemma \ --lora_rank 64 \ --output_dir ./output \ --num_epochs 3 \ --batch_size 4 \ --load_in_4bit@article{abdulmumin2026afriscience, title = {AfriScience-MT: Towards Decolonizing Science in Africa through Text Translation}, author = {Abdulmumin, Idris and Gwadabe, Tajuddeen and Muhammad, Shamsuddeen Hassan and Adelani, David Ifeoluwa and Khalo, Nomonde and Ahmad, Ibrahim Said and Modupe, Abiodun and Mumm, Anina and Biyela, Sibusiso and Rabie, Michelle and Havemann, Johanna and Rei, Marek and Abbott, Jade and Marivate, Vukosi}, journal = {arXiv preprint arXiv:2605.29741}, year = {2026}, url = {https://arxiv.org/abs/2605.29741}}