Text Generation
PEFT
Safetensors
English
lora
finance
sentiment-analysis
financial-phrasebank
finben
negative-result
Eval Results (legacy)
Instructions to use Algorex/FinDan-lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use Algorex/FinDan-lora with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("ChanceFocus/finma-7b-nlp") model = PeftModel.from_pretrained(base_model, "Algorex/FinDan-lora") - Notebooks
- Google Colab
- Kaggle
Download recipe.json from Algorex/FinDan-lora: direct link, hf CLI and curl.
- Browser
- Download file 2.2 kB
-
https://huggingface.co/Algorex/FinDan-lora/resolve/main/recipe.json
- Command line
-
hf download hf://Algorex/FinDan-lora/recipe.json
-
curl -L -o recipe.json https://huggingface.co/Algorex/FinDan-lora/resolve/main/recipe.json
2.2 kB
| { | |
| "schema_version": 1, | |
| "created_at_utc": "2026-11-07T07:27:37.369019+00:00", | |
| "name": "finma-finma_wrapper-r16-a32-attn_mlp-lr5e-5-e2-cosine-s3407-trainval", | |
| "track": "finma_derived", | |
| "lane": "finma_wrapper", | |
| "base_model": "ChanceFocus/finma-7b-nlp", | |
| "base_revision": "94be8e68ca0b5fd90fbfb8cb87b130ea9e423270", | |
| "loading": "4-bit NF4, compute float16, torch_dtype float16 (= protocol baseline.model_args)", | |
| "hyperparameters": { | |
| "rank": 16, | |
| "alpha": 32, | |
| "targets": "attn_mlp", | |
| "target_modules": [ | |
| "q_proj", | |
| "k_proj", | |
| "v_proj", | |
| "o_proj", | |
| "gate_proj", | |
| "up_proj", | |
| "down_proj" | |
| ], | |
| "dropout": 0.05, | |
| "lr": 5e-05, | |
| "epochs": 2.0, | |
| "scheduler": "cosine", | |
| "warmup_ratio": 0.03, | |
| "micro_batch": 8, | |
| "grad_accum": 2, | |
| "effective_batch": 16, | |
| "weight_decay": 0.0, | |
| "optim": "adamw_torch", | |
| "fp16_amp": true, | |
| "max_steps": null | |
| }, | |
| "seed": 3407, | |
| "data": "train_validation", | |
| "data_files": { | |
| "train": { | |
| "path": "finben_train.csv", | |
| "sha256": "d9a292c7fe921b2a8f632e7847497aaf8c60ec8d7eca4e2494b78f50447a11b2", | |
| "rows": 3100 | |
| }, | |
| "validation": { | |
| "path": "finben_validation.csv", | |
| "sha256": "b79adaac4c3d8b93f3a6720fedfdb5f840e7a5c626d8d72c7067d5fbd0c1ee3b", | |
| "rows": 776 | |
| } | |
| }, | |
| "train_csv": "finben_train.csv", | |
| "train_sha256": "d9a292c7fe921b2a8f632e7847497aaf8c60ec8d7eca4e2494b78f50447a11b2", | |
| "train_rows": 3876, | |
| "protocol_sha256": "ef24b51d26b7e0b3c58a4c068abfad7368175a1b32cc834b03e970f851083311", | |
| "trainable_parameters": 39976960, | |
| "global_steps": 484, | |
| "final_train_loss": 0.05175297841244985, | |
| "seconds": 2518.9846563339233, | |
| "seconds_per_step": 5.20451375275604, | |
| "peak_vram_gib": 6.870232582092285, | |
| "packages": { | |
| "torch": "2.7.1+cu128", | |
| "transformers": "4.40.2", | |
| "tokenizers": "0.19.1", | |
| "peft": "0.9.0", | |
| "accelerate": "0.30.1", | |
| "bitsandbytes": "0.46.1", | |
| "sentencepiece": "0.2.3" | |
| }, | |
| "smoke": false, | |
| "note": "Copy of the adapter's recipe.json with local directory paths reduced to file names; byte-identical weights." | |
| } |