Upload model
Browse files- README.md +15 -22
- adapter_config.json +8 -2
README.md
CHANGED
|
@@ -1,28 +1,21 @@
|
|
| 1 |
---
|
| 2 |
-
|
| 3 |
-
language:
|
| 4 |
-
- es
|
| 5 |
-
library_name: transformers
|
| 6 |
-
tags:
|
| 7 |
-
- falcon
|
| 8 |
-
- alpaca
|
| 9 |
-
- Transformers
|
| 10 |
-
- gpt
|
| 11 |
-
- PyTorch
|
| 12 |
-
- llm
|
| 13 |
-
- llm spanish
|
| 14 |
-
pipeline_tag: text-generation
|
| 15 |
-
datasets:
|
| 16 |
-
- bertin-project/alpaca-spanish
|
| 17 |
---
|
|
|
|
| 18 |
|
| 19 |
-
<strong><span style="font-size: larger;">FALCON 7B Spanish Fine-tuned 8bit 🤗</span></strong>
|
| 20 |
|
| 21 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
|
| 23 |
-
The dataset is a translation to Spanish of alpaca_data_cleaned.json (a clean version of the Alpaca dataset made at Stanford) using OpenAI's gpt-3.5-turbo model. This translation was made by bertin-project. It was translated using a full-sample prompt instead of per strings, which resulted in more coherent tuples of (instruction, input, output).
|
| 24 |
-
Dataset link: [here](https://huggingface.co/datasets/bertin-project/alpaca-spanish)
|
| 25 |
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
To fine-tune the FALCON-7B model we used the [following code](https://github.com/AdrianBZG/LLM-distributed-finetune) to run it on a distributed cluster on AWS. You are free to use such code as a fingerprint to finetune any model as you please, as it is easily customizable.
|
|
|
|
| 1 |
---
|
| 2 |
+
library_name: peft
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
| 4 |
+
## Training procedure
|
| 5 |
|
|
|
|
| 6 |
|
| 7 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 8 |
+
- quant_method: bitsandbytes
|
| 9 |
+
- load_in_8bit: True
|
| 10 |
+
- load_in_4bit: False
|
| 11 |
+
- llm_int8_threshold: 6.0
|
| 12 |
+
- llm_int8_skip_modules: None
|
| 13 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 14 |
+
- llm_int8_has_fp16_weight: False
|
| 15 |
+
- bnb_4bit_quant_type: fp4
|
| 16 |
+
- bnb_4bit_use_double_quant: False
|
| 17 |
+
- bnb_4bit_compute_dtype: float32
|
| 18 |
+
### Framework versions
|
| 19 |
|
|
|
|
|
|
|
| 20 |
|
| 21 |
+
- PEFT 0.6.0.dev0
|
|
|
|
|
|
adapter_config.json
CHANGED
|
@@ -1,14 +1,20 @@
|
|
| 1 |
{
|
|
|
|
|
|
|
| 2 |
"base_model_name_or_path": "tiiuae/falcon-7b",
|
| 3 |
"bias": "none",
|
| 4 |
"fan_in_fan_out": false,
|
| 5 |
"inference_mode": true,
|
| 6 |
"init_lora_weights": true,
|
| 7 |
-
"
|
|
|
|
|
|
|
| 8 |
"lora_dropout": 0.05,
|
| 9 |
"modules_to_save": null,
|
| 10 |
"peft_type": "LORA",
|
| 11 |
-
"r":
|
|
|
|
|
|
|
| 12 |
"target_modules": [
|
| 13 |
"query_key_value"
|
| 14 |
],
|
|
|
|
| 1 |
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": null,
|
| 4 |
"base_model_name_or_path": "tiiuae/falcon-7b",
|
| 5 |
"bias": "none",
|
| 6 |
"fan_in_fan_out": false,
|
| 7 |
"inference_mode": true,
|
| 8 |
"init_lora_weights": true,
|
| 9 |
+
"layers_pattern": null,
|
| 10 |
+
"layers_to_transform": null,
|
| 11 |
+
"lora_alpha": 32,
|
| 12 |
"lora_dropout": 0.05,
|
| 13 |
"modules_to_save": null,
|
| 14 |
"peft_type": "LORA",
|
| 15 |
+
"r": 16,
|
| 16 |
+
"rank_pattern": {},
|
| 17 |
+
"revision": null,
|
| 18 |
"target_modules": [
|
| 19 |
"query_key_value"
|
| 20 |
],
|