Update README.md
Browse files
README.md
CHANGED
@@ -23,20 +23,8 @@ model_name = "jwnder/core42_jais-13b-bnb-4bit"
|
|
23 |
import warnings
|
24 |
warnings.filterwarnings('ignore')
|
25 |
|
26 |
-
|
27 |
-
|
28 |
-
bnb_4bit_quant_type="nf4",
|
29 |
-
bnb_4bit_compute_dtype=torch.bfloat16,
|
30 |
-
bnb_4bit_use_double_quant=True,
|
31 |
-
llm_int8_enable_fp32_cpu_offload=True
|
32 |
-
)
|
33 |
-
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
|
34 |
-
model = AutoModelForCausalLM.from_pretrained(
|
35 |
-
model_name,
|
36 |
-
quantization_config=bnb_config,
|
37 |
-
device_map="auto",
|
38 |
-
trust_remote_code=True
|
39 |
-
)
|
40 |
|
41 |
inputs = tokenizer("Testing LLM!", return_tensors="pt")
|
42 |
start = datetime.now()
|
|
|
23 |
import warnings
|
24 |
warnings.filterwarnings('ignore')
|
25 |
|
26 |
+
tokenizer = AutoTokenizer.from_pretrained(model_input_folder, trust_remote_code=True)
|
27 |
+
model = AutoModelForCausalLM.from_pretrained(model_input_folder, trust_remote_code=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
28 |
|
29 |
inputs = tokenizer("Testing LLM!", return_tensors="pt")
|
30 |
start = datetime.now()
|