Text Generation
PEFT
Safetensors
English
Spanish
llama4_text
agriculture
climate
crop-calendar
evidence-grounding
bilingual
lora
autoscientist
adaption
conversational
4-bit precision
bitsandbytes
Instructions to use MarianaCodebase/AgroVeritas-Scout-17B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use MarianaCodebase/AgroVeritas-Scout-17B with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("togethercomputer/Llama-4-Scout-17B-16E-Instruct_bnb_4bit") model = PeftModel.from_pretrained(base_model, "MarianaCodebase/AgroVeritas-Scout-17B") - Notebooks
- Google Colab
- Kaggle
Download autoscientist_config.json from MarianaCodebase/AgroVeritas-Scout-17B: direct link, hf CLI and curl.
- Browser
- Download file 896 Bytes
-
https://huggingface.co/MarianaCodebase/AgroVeritas-Scout-17B/resolve/main/autoscientist_config.json
- Command line
-
hf download hf://MarianaCodebase/AgroVeritas-Scout-17B/autoscientist_config.json
-
curl -L -o autoscientist_config.json https://huggingface.co/MarianaCodebase/AgroVeritas-Scout-17B/resolve/main/autoscientist_config.json
896 Bytes
| { | |
| "model": "meta-llama/Llama-4-Scout-17B-16E-Instruct", | |
| "exported_base_reference": "togethercomputer/Llama-4-Scout-17B-16E-Instruct_bnb_4bit", | |
| "winning_autoscientist_model_id": "adaption_llama_4_scout_17b_16_agri_evidence_qa_158a7fd8", | |
| "hyperparams": { | |
| "base_model_size": "109B", | |
| "n_epochs": 3, | |
| "batch_size": "max", | |
| "learning_rate": 0.0001, | |
| "lora": true, | |
| "lora_r": 64, | |
| "lora_alpha": 128, | |
| "lora_dropout": 0, | |
| "lora_trainable_modules": "q_proj,k_proj,v_proj,o_proj,shared_expert.gate_proj,shared_expert.up_proj,shared_expert.down_proj,feed_forward.gate_proj,feed_forward.up_proj,feed_forward.down_proj", | |
| "lr_scheduler_type": "cosine", | |
| "min_lr_ratio": 0.1, | |
| "scheduler_num_cycles": 0.5, | |
| "warmup_ratio": 0.1, | |
| "max_grad_norm": 1, | |
| "weight_decay": 0.01, | |
| "n_evals": 5, | |
| "training_method": "sft", | |
| "train_on_inputs": false | |
| } | |
| } | |