Update README.md
Browse files
README.md
CHANGED
@@ -18,11 +18,39 @@ This model is a fine-tuned version of [databricks/dolly-v2-3b](https://huggingfa
|
|
18 |
|
19 |
## Model description
|
20 |
|
21 |
-
This is a
|
22 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
23 |
## Intended uses & limitations
|
24 |
|
25 |
-
|
|
|
|
|
|
|
26 |
|
27 |
## Training and evaluation data
|
28 |
|
|
|
18 |
|
19 |
## Model description
|
20 |
|
21 |
+
This is a PEFT model, hence the model file and the config files are
|
22 |
+
* adapter_model.bin
|
23 |
+
* adapter_config.bin
|
24 |
+
|
25 |
+
This fined-tuned model was created with the following bitsandbytes config<br>
|
26 |
+
|
27 |
+
BitsAndBytesConfig(load_in_8bit = True,
|
28 |
+
bnb_4bit_quant_type = 'nf4',
|
29 |
+
bnb_4bit_compute_type = torch.bfloat16,
|
30 |
+
bnb_4bit_use_double_quant = True)
|
31 |
+
|
32 |
+
The peft_config is as follows:
|
33 |
+
|
34 |
+
peft_config = LoraConfig(
|
35 |
+
lora_alpha=16,
|
36 |
+
lora_dropout = 0.1,
|
37 |
+
r = 64,
|
38 |
+
bias = "none",
|
39 |
+
task_type = "CAUSAL_LM",
|
40 |
+
target_modules = [
|
41 |
+
'query_key_value',
|
42 |
+
'dense',
|
43 |
+
'dense_h_to_4h',
|
44 |
+
'dense_4h_to_h'
|
45 |
+
]
|
46 |
+
)
|
47 |
+
</br>
|
48 |
## Intended uses & limitations
|
49 |
|
50 |
+
Model is intended for fair use only.
|
51 |
+
|
52 |
+
|
53 |
+
|
54 |
|
55 |
## Training and evaluation data
|
56 |
|