VitaliiVrublevskyi
commited on
Commit
·
e043c8c
1
Parent(s):
a5a17e2
update model card README.md
Browse files
README.md
CHANGED
@@ -10,7 +10,6 @@ metrics:
|
|
10 |
model-index:
|
11 |
- name: Llama-2-7b-hf-finetuned-mrpc-v5
|
12 |
results: []
|
13 |
-
library_name: peft
|
14 |
---
|
15 |
|
16 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
@@ -20,9 +19,9 @@ should probably proofread and complete it, then remove this comment. -->
|
|
20 |
|
21 |
This model is a fine-tuned version of [meta-llama/Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) on the glue dataset.
|
22 |
It achieves the following results on the evaluation set:
|
23 |
-
- Loss: 0.
|
24 |
-
- Accuracy: 0.
|
25 |
-
- F1: 0.
|
26 |
|
27 |
## Model description
|
28 |
|
@@ -38,17 +37,6 @@ More information needed
|
|
38 |
|
39 |
## Training procedure
|
40 |
|
41 |
-
|
42 |
-
The following `bitsandbytes` quantization config was used during training:
|
43 |
-
- load_in_8bit: True
|
44 |
-
- load_in_4bit: False
|
45 |
-
- llm_int8_threshold: 6.0
|
46 |
-
- llm_int8_skip_modules: None
|
47 |
-
- llm_int8_enable_fp32_cpu_offload: False
|
48 |
-
- llm_int8_has_fp16_weight: False
|
49 |
-
- bnb_4bit_quant_type: fp4
|
50 |
-
- bnb_4bit_use_double_quant: False
|
51 |
-
- bnb_4bit_compute_dtype: float32
|
52 |
### Training hyperparameters
|
53 |
|
54 |
The following hyperparameters were used during training:
|
@@ -58,26 +46,28 @@ The following hyperparameters were used during training:
|
|
58 |
- seed: 42
|
59 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
60 |
- lr_scheduler_type: linear
|
61 |
-
- num_epochs:
|
62 |
|
63 |
### Training results
|
64 |
|
65 |
-
| Training Loss | Epoch | Step
|
66 |
-
|
67 |
-
| 0.733 | 1.0 | 917
|
68 |
-
| 0.6103 | 2.0 | 1834
|
69 |
-
| 0.508 | 3.0 | 2751
|
70 |
-
| 0.5354 | 4.0 | 3668
|
71 |
-
| 0.4246 | 5.0 | 4585
|
72 |
-
| 0.4295 | 6.0 | 5502
|
73 |
-
| 0.4473 | 7.0 | 6419
|
74 |
-
| 0.419 | 8.0 | 7336
|
75 |
-
| 0.3876 | 9.0 | 8253
|
|
|
|
|
|
|
76 |
|
77 |
|
78 |
### Framework versions
|
79 |
|
80 |
-
- PEFT 0.4.0
|
81 |
- Transformers 4.31.0
|
82 |
- Pytorch 2.0.1+cu118
|
83 |
- Datasets 2.14.5
|
|
|
10 |
model-index:
|
11 |
- name: Llama-2-7b-hf-finetuned-mrpc-v5
|
12 |
results: []
|
|
|
13 |
---
|
14 |
|
15 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
19 |
|
20 |
This model is a fine-tuned version of [meta-llama/Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) on the glue dataset.
|
21 |
It achieves the following results on the evaluation set:
|
22 |
+
- Loss: 0.6070
|
23 |
+
- Accuracy: 0.8480
|
24 |
+
- F1: 0.8916
|
25 |
|
26 |
## Model description
|
27 |
|
|
|
37 |
|
38 |
## Training procedure
|
39 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
40 |
### Training hyperparameters
|
41 |
|
42 |
The following hyperparameters were used during training:
|
|
|
46 |
- seed: 42
|
47 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
48 |
- lr_scheduler_type: linear
|
49 |
+
- num_epochs: 12
|
50 |
|
51 |
### Training results
|
52 |
|
53 |
+
| Training Loss | Epoch | Step | Accuracy | F1 | Validation Loss |
|
54 |
+
|:-------------:|:-----:|:-----:|:--------:|:------:|:---------------:|
|
55 |
+
| 0.733 | 1.0 | 917 | 0.6912 | 0.7974 | 0.6016 |
|
56 |
+
| 0.6103 | 2.0 | 1834 | 0.7402 | 0.8339 | 0.5650 |
|
57 |
+
| 0.508 | 3.0 | 2751 | 0.7525 | 0.8358 | 0.5246 |
|
58 |
+
| 0.5354 | 4.0 | 3668 | 0.7794 | 0.8529 | 0.5318 |
|
59 |
+
| 0.4246 | 5.0 | 4585 | 0.7843 | 0.8508 | 0.5279 |
|
60 |
+
| 0.4295 | 6.0 | 5502 | 0.7966 | 0.8591 | 0.5248 |
|
61 |
+
| 0.4473 | 7.0 | 6419 | 0.8162 | 0.8696 | 0.5169 |
|
62 |
+
| 0.419 | 8.0 | 7336 | 0.8260 | 0.8778 | 0.5552 |
|
63 |
+
| 0.3876 | 9.0 | 8253 | 0.8284 | 0.8776 | 0.5514 |
|
64 |
+
| 0.42 | 10.0 | 9170 | 0.5576 | 0.8407 | 0.8862 |
|
65 |
+
| 0.3678 | 11.0 | 10087 | 0.6212 | 0.8480 | 0.8927 |
|
66 |
+
| 0.3453 | 12.0 | 11004 | 0.6070 | 0.8480 | 0.8916 |
|
67 |
|
68 |
|
69 |
### Framework versions
|
70 |
|
|
|
71 |
- Transformers 4.31.0
|
72 |
- Pytorch 2.0.1+cu118
|
73 |
- Datasets 2.14.5
|