address comments

h-guo18 · yeyu-nvidia · h-guo18 · commit 70355652188f · 2025-09-17T13:34:05.000-07:00
Co-authored-by: yeyu-nvidia &lt;yeyu@nvidia.com&gt;
Signed-off-by: h-guo18 &lt;67671475+h-guo18@users.noreply.github.com&gt;
diff --git a/examples/speculative_decoding/example.ipynb b/examples/speculative_decoding/example.ipynb
@@ -77,7 +77,7 @@
    "metadata": {},
    "source": [
     "## Train Draft module On Daring-Anteater\n",
-    "We will fine-tune the draft head on the Daring-Anteater dataset using the standard Hugging Face Trainer. Note that only the draft model's weights are updated during this process; the original target model remains unchanged. After training, our speculative decoding model will be ready for export and deployment. Note that the time to train will be significantly dependent on the epochs (default=4) and the hardware being used."
+    "We will fine-tune the draft module on the Daring-Anteater dataset using the standard Hugging Face Trainer. Note that only the draft module's weights are updated during this process; the original target model remains frozen. After training, our speculative decoding model will be ready for export and deployment. Note that the time to train will be significantly dependent on the epochs (default=4) and the hardware being used."
    ]
   },
   {

Original file line number	Diff line number	Diff line change
`@@ -77,7 +77,7 @@`
`77`	`77`	`"metadata": {},`
`78`	`78`	`"source": [`
`79`	`79`	`"## Train Draft module On Daring-Anteater\n",`
`80`		`- "We will fine-tune the draft head on the Daring-Anteater dataset using the standard Hugging Face Trainer. Note that only the draft model's weights are updated during this process; the original target model remains unchanged. After training, our speculative decoding model will be ready for export and deployment. Note that the time to train will be significantly dependent on the epochs (default=4) and the hardware being used."`
	`80`	`+ "We will fine-tune the draft module on the Daring-Anteater dataset using the standard Hugging Face Trainer. Note that only the draft module's weights are updated during this process; the original target model remains frozen. After training, our speculative decoding model will be ready for export and deployment. Note that the time to train will be significantly dependent on the epochs (default=4) and the hardware being used."`
`81`	`81`	`]`
`82`	`82`	`},`
`83`	`83`	`{`