Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,27 @@
|
|
1 |
---
|
2 |
license: apache-2.0
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: apache-2.0
|
3 |
---
|
4 |
+
note : use original open llama tokenizer
|
5 |
+
|
6 |
+
#model_path = 'openlm-research/open_llama_3b'
|
7 |
+
model_path = 'ruwan/open-llama-sharded-1GB-7B-alpaca-vmware'
|
8 |
+
# model_path = 'openlm-research/open_llama_13b_600bt'
|
9 |
+
|
10 |
+
tokenizer = LlamaTokenizer.from_pretrained("openlm-research/open_llama_7b")
|
11 |
+
model = LlamaForCausalLM.from_pretrained(
|
12 |
+
model_path, torch_dtype=torch.float16, device_map='auto',load_in_8bit=True
|
13 |
+
)
|
14 |
+
|
15 |
+
|
16 |
+
prompt_template = "Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n### Instruction:\n{instruction}\n\n### Response:"
|
17 |
+
|
18 |
+
prompt= 'Explain in simple terms how the attention mechanism of a transformer model works'
|
19 |
+
|
20 |
+
|
21 |
+
inputt = prompt_template.format(instruction= prompt)
|
22 |
+
input_ids = tokenizer(inputt, return_tensors="pt").input_ids.to("cuda")
|
23 |
+
|
24 |
+
output1 = model.generate(input_ids, max_length=512)
|
25 |
+
input_length = input_ids.shape[1]
|
26 |
+
output1 = output1[:, input_length:]
|
27 |
+
output= tokenizer.decode(output1[0])
|