visheratin Abhaykoul commited on
Commit
8965660
1 Parent(s): 3f4bd65

Update README.md (#1)

Browse files

- Update README.md (1db2e3a5b427e553bab79f6707da43353aaa6f16)


Co-authored-by: HelpingAI <Abhaykoul@users.noreply.huggingface.co>

Files changed (1) hide show
  1. README.md +8 -7
README.md CHANGED
@@ -8,6 +8,7 @@ tags:
8
  - llava
9
  - phi
10
  license: mit
 
11
  ---
12
 
13
  # LLaVA-3b
@@ -40,13 +41,13 @@ You are Dolphin, a helpful AI assistant.<|im_end|>
40
 
41
  **Install dependencies**
42
 
43
- ```
44
  !pip install -q open_clip_torch timm einops
45
  ```
46
 
47
  **Download modeling files**
48
 
49
- ```
50
  from huggingface_hub import hf_hub_download
51
 
52
  hf_hub_download(repo_id="visheratin/LLaVA-3b", filename="configuration_llava.py", local_dir="./", force_download=True)
@@ -58,7 +59,7 @@ hf_hub_download(repo_id="visheratin/LLaVA-3b", filename="processing_llava.py", l
58
 
59
  **Create a model**
60
 
61
- ```
62
  from modeling_llava import LlavaForConditionalGeneration
63
  import torch
64
 
@@ -68,7 +69,7 @@ model = model.to("cuda")
68
 
69
  **Create processors**
70
 
71
- ```
72
  from transformers import AutoTokenizer
73
  from processing_llava import LlavaProcessor, OpenCLIPImageProcessor
74
 
@@ -79,7 +80,7 @@ processor = LlavaProcessor(image_processor, tokenizer)
79
 
80
  **Set image and text**
81
 
82
- ```
83
  from PIL import Image
84
  import requests
85
 
@@ -99,7 +100,7 @@ Describe the image.<|im_end|>
99
 
100
  **Process inputs**
101
 
102
- ```
103
  inputs = processor(prompt, raw_image, model, return_tensors='pt')
104
 
105
  inputs['input_ids'] = inputs['input_ids'].to(model.device)
@@ -108,7 +109,7 @@ inputs['attention_mask'] = inputs['attention_mask'].to(model.device)
108
 
109
  **Generate the data**
110
 
111
- ```
112
  output = model.generate(**inputs, max_new_tokens=200, do_sample=True, top_p=0.5, temperature=1.2, eos_token_id=tokenizer.eos_token_id)
113
  ```
114
 
 
8
  - llava
9
  - phi
10
  license: mit
11
+ library_name: transformers
12
  ---
13
 
14
  # LLaVA-3b
 
41
 
42
  **Install dependencies**
43
 
44
+ ```bash
45
  !pip install -q open_clip_torch timm einops
46
  ```
47
 
48
  **Download modeling files**
49
 
50
+ ```python
51
  from huggingface_hub import hf_hub_download
52
 
53
  hf_hub_download(repo_id="visheratin/LLaVA-3b", filename="configuration_llava.py", local_dir="./", force_download=True)
 
59
 
60
  **Create a model**
61
 
62
+ ```python
63
  from modeling_llava import LlavaForConditionalGeneration
64
  import torch
65
 
 
69
 
70
  **Create processors**
71
 
72
+ ```python
73
  from transformers import AutoTokenizer
74
  from processing_llava import LlavaProcessor, OpenCLIPImageProcessor
75
 
 
80
 
81
  **Set image and text**
82
 
83
+ ```python
84
  from PIL import Image
85
  import requests
86
 
 
100
 
101
  **Process inputs**
102
 
103
+ ```python
104
  inputs = processor(prompt, raw_image, model, return_tensors='pt')
105
 
106
  inputs['input_ids'] = inputs['input_ids'].to(model.device)
 
109
 
110
  **Generate the data**
111
 
112
+ ```python
113
  output = model.generate(**inputs, max_new_tokens=200, do_sample=True, top_p=0.5, temperature=1.2, eos_token_id=tokenizer.eos_token_id)
114
  ```
115