Spaces:
Running
Running
jadehardouin
commited on
Commit
•
9793af4
1
Parent(s):
1c2b775
Update models.py
Browse files
models.py
CHANGED
@@ -112,7 +112,7 @@ class OpenAIModel(BaseTCOModel):
|
|
112 |
|
113 |
self.labor = gr.Number(0, visible=False,
|
114 |
label="($) Labor cost per month",
|
115 |
-
info="This is
|
116 |
interactive=True
|
117 |
)
|
118 |
|
@@ -144,8 +144,9 @@ class OpenSourceLlama2Model(BaseTCOModel):
|
|
144 |
visible=False,
|
145 |
label="Instance of VM with GPU",
|
146 |
)
|
147 |
-
self.vm_cost_per_hour = gr.Number(
|
148 |
interactive=False, visible=False)
|
|
|
149 |
self.input_tokens_cost_per_second = gr.Number(0.00052, visible=False,
|
150 |
label="($) Price/1K input prompt tokens",
|
151 |
interactive=False
|
@@ -154,18 +155,17 @@ class OpenSourceLlama2Model(BaseTCOModel):
|
|
154 |
label="($) Price/1K output prompt tokens",
|
155 |
interactive=False
|
156 |
)
|
157 |
-
self.
|
|
|
|
|
|
|
|
|
158 |
|
159 |
self.labor = gr.Number(10000, visible=False,
|
160 |
label="($) Labor cost per month",
|
161 |
-
info="This is
|
162 |
interactive=True
|
163 |
)
|
164 |
-
|
165 |
-
# self.used = gr.Slider(minimum=0.01, value=30., step=0.01, label="% used",
|
166 |
-
# info="Percentage of time the GPU is used",
|
167 |
-
# interactive=True,
|
168 |
-
# visible=False)
|
169 |
|
170 |
def compute_cost_per_token(self, input_tokens_cost_per_second, output_tokens_cost_per_second, labor):
|
171 |
cost_per_input_token = (input_tokens_cost_per_second / 1000)
|
@@ -201,7 +201,7 @@ class CohereModel(BaseTCOModel):
|
|
201 |
|
202 |
self.labor = gr.Number(0, visible=False,
|
203 |
label="($) Labor cost per month",
|
204 |
-
info="This is
|
205 |
interactive=True
|
206 |
)
|
207 |
|
@@ -254,7 +254,7 @@ class ModelPage:
|
|
254 |
for model in self.models:
|
255 |
if model.get_name() == name:
|
256 |
output+= [gr.update(visible=True)] * len(model.get_components())
|
257 |
-
# Set use_case
|
258 |
model.use_case = use_case
|
259 |
else:
|
260 |
output+= [gr.update(visible=False)] * len(model.get_components())
|
|
|
112 |
|
113 |
self.labor = gr.Number(0, visible=False,
|
114 |
label="($) Labor cost per month",
|
115 |
+
info="This is an estimate of the labor cost of the AI engineer in charge of deploying the model",
|
116 |
interactive=True
|
117 |
)
|
118 |
|
|
|
144 |
visible=False,
|
145 |
label="Instance of VM with GPU",
|
146 |
)
|
147 |
+
self.vm_cost_per_hour = gr.Number(4.42, label="Instance cost ($) per hour",
|
148 |
interactive=False, visible=False)
|
149 |
+
self.info_vm = gr.Markdown("This price above is from [CoreWeave's pricing web page](https://www.coreweave.com/gpu-cloud-pricing)", interactive=False, visible=False)
|
150 |
self.input_tokens_cost_per_second = gr.Number(0.00052, visible=False,
|
151 |
label="($) Price/1K input prompt tokens",
|
152 |
interactive=False
|
|
|
155 |
label="($) Price/1K output prompt tokens",
|
156 |
interactive=False
|
157 |
)
|
158 |
+
self.source = gr.Markdown("""<span style="font-size: 16px; font-weight: 600; color: #212529;">Source</span>""")
|
159 |
+
self.info = gr.Markdown("The cost per input and output tokens values above are from [these benchmark results](https://www.cursor.so/blog/llama-inference#user-content-fn-llama-paper)",
|
160 |
+
label="Source",
|
161 |
+
interactive=False,
|
162 |
+
visible=False)
|
163 |
|
164 |
self.labor = gr.Number(10000, visible=False,
|
165 |
label="($) Labor cost per month",
|
166 |
+
info="This is an estimate of the labor cost of the AI engineer in charge of deploying the model",
|
167 |
interactive=True
|
168 |
)
|
|
|
|
|
|
|
|
|
|
|
169 |
|
170 |
def compute_cost_per_token(self, input_tokens_cost_per_second, output_tokens_cost_per_second, labor):
|
171 |
cost_per_input_token = (input_tokens_cost_per_second / 1000)
|
|
|
201 |
|
202 |
self.labor = gr.Number(0, visible=False,
|
203 |
label="($) Labor cost per month",
|
204 |
+
info="This is an estimate of the labor cost of the AI engineer in charge of deploying the model",
|
205 |
interactive=True
|
206 |
)
|
207 |
|
|
|
254 |
for model in self.models:
|
255 |
if model.get_name() == name:
|
256 |
output+= [gr.update(visible=True)] * len(model.get_components())
|
257 |
+
# Set use_case value in the model
|
258 |
model.use_case = use_case
|
259 |
else:
|
260 |
output+= [gr.update(visible=False)] * len(model.get_components())
|