Text Generation
Transformers
Safetensors
llama
code
granite
conversational
Eval Results
text-generation-inference
mayank-mishra commited on
Commit
596a400
1 Parent(s): 9f8823b

update example

Browse files
Files changed (1) hide show
  1. README.md +8 -5
README.md CHANGED
@@ -226,25 +226,28 @@ This is a simple example of how to use **Granite-8B-Code-Instruct** model.
226
  import torch
227
  from transformers import AutoModelForCausalLM, AutoTokenizer
228
  device = "cuda" # or "cpu"
229
- model_path = "ibm-granite/granite-8b-code-instruct"
230
  tokenizer = AutoTokenizer.from_pretrained(model_path)
231
  # drop device_map if running on CPU
232
  model = AutoModelForCausalLM.from_pretrained(model_path, device_map=device)
233
  model.eval()
234
  # change input text as desired
235
- input_text = "Write a code to find the maximum value in a list of numbers. The list can contain both positive and negative numbers, and the maximum value can be either a positive or negative number."
 
 
 
236
  # tokenize the text
237
- input_tokens = tokenizer(input_text, return_tensors="pt")
238
  # transfer tokenized inputs to the device
239
  for i in input_tokens:
240
  input_tokens[i] = input_tokens[i].to(device)
241
  # generate output tokens
242
- output = model.generate(**input_tokens)
243
  # decode output tokens into text
244
  output = tokenizer.batch_decode(output)
245
  # loop over the batch to print, in this example the batch size is 1
246
  for i in output:
247
- print(output)
248
  ```
249
 
250
  <!-- TO DO: Check this part -->
 
226
  import torch
227
  from transformers import AutoModelForCausalLM, AutoTokenizer
228
  device = "cuda" # or "cpu"
229
+ model_path = "granite-8b-code-instruct"
230
  tokenizer = AutoTokenizer.from_pretrained(model_path)
231
  # drop device_map if running on CPU
232
  model = AutoModelForCausalLM.from_pretrained(model_path, device_map=device)
233
  model.eval()
234
  # change input text as desired
235
+ chat = [
236
+ { "role": "user", "content": "Write a code to find the maximum value in a list of numbers." },
237
+ ]
238
+ chat = tokenizer.apply_chat_template(chat, tokenize=False, add_generation_prompt=True)
239
  # tokenize the text
240
+ input_tokens = tokenizer(chat, return_tensors="pt")
241
  # transfer tokenized inputs to the device
242
  for i in input_tokens:
243
  input_tokens[i] = input_tokens[i].to(device)
244
  # generate output tokens
245
+ output = model.generate(**input_tokens, max_new_tokens=100)
246
  # decode output tokens into text
247
  output = tokenizer.batch_decode(output)
248
  # loop over the batch to print, in this example the batch size is 1
249
  for i in output:
250
+ print(i)
251
  ```
252
 
253
  <!-- TO DO: Check this part -->