Spaces:
Running
Running
Better caching
Browse files
app.py
CHANGED
@@ -62,13 +62,14 @@ if metric_name == "KL divergence":
|
|
62 |
tokenizer = st.cache_resource(AutoTokenizer.from_pretrained, show_spinner=False)(model_name)
|
63 |
model = st.cache_resource(AutoModelForCausalLM.from_pretrained, show_spinner=False)(model_name)
|
64 |
|
|
|
|
|
|
|
|
|
65 |
@st.cache_data(show_spinner=False)
|
66 |
def run_context_length_probing(model_name, text, window_len):
|
67 |
assert model.name_or_path == model_name
|
68 |
-
|
69 |
-
inputs = tokenizer([text])
|
70 |
-
[input_ids] = inputs["input_ids"]
|
71 |
-
window_len = min(window_len, len(input_ids))
|
72 |
|
73 |
inputs_sliding = get_windows_batched(
|
74 |
inputs,
|
@@ -89,9 +90,9 @@ def run_context_length_probing(model_name, text, window_len):
|
|
89 |
scores /= scores.abs().max(dim=1, keepdim=True).values + 1e-9
|
90 |
scores = scores.to(torch.float16)
|
91 |
|
92 |
-
return
|
93 |
|
94 |
-
|
95 |
model_name=model_name,
|
96 |
text=text,
|
97 |
window_len=window_len
|
|
|
62 |
tokenizer = st.cache_resource(AutoTokenizer.from_pretrained, show_spinner=False)(model_name)
|
63 |
model = st.cache_resource(AutoModelForCausalLM.from_pretrained, show_spinner=False)(model_name)
|
64 |
|
65 |
+
inputs = tokenizer([text])
|
66 |
+
[input_ids] = inputs["input_ids"]
|
67 |
+
window_len = min(window_len, len(input_ids))
|
68 |
+
|
69 |
@st.cache_data(show_spinner=False)
|
70 |
def run_context_length_probing(model_name, text, window_len):
|
71 |
assert model.name_or_path == model_name
|
72 |
+
del text # needed as a cache key but for the computation we access inputs directly
|
|
|
|
|
|
|
73 |
|
74 |
inputs_sliding = get_windows_batched(
|
75 |
inputs,
|
|
|
90 |
scores /= scores.abs().max(dim=1, keepdim=True).values + 1e-9
|
91 |
scores = scores.to(torch.float16)
|
92 |
|
93 |
+
return scores
|
94 |
|
95 |
+
scores = run_context_length_probing(
|
96 |
model_name=model_name,
|
97 |
text=text,
|
98 |
window_len=window_len
|