chunk size 1MB for 2-3x throughput improvement

app.py

@@ -129,7 +129,7 @@ def stream_from_hf(dataset, file_path, wfile, range_start=0, range_end=None):
     resp = urllib.request.urlopen(req, timeout=300)
     total = int(resp.headers.get("Content-Length", 0))
     while True:
-        chunk = resp.read(65536)
+        chunk = resp.read(1048576)
         if not chunk:
             break
         wfile.write(chunk)
@@ -299,7 +299,7 @@ class Handler(http.server.BaseHTTPRequestHandler):
             f.seek(start)
             remaining = length
             while remaining > 0:
-                chunk_size = min(65536, remaining)
+                chunk_size = min(1048576, remaining)
                 chunk = f.read(chunk_size)
                 if not chunk:
                     break
@@ -352,7 +352,7 @@ class Handler(http.server.BaseHTTPRequestHandler):
         try:
             tmpf = open(temp, "wb") if temp else None
             while True:
-                chunk = hf_resp.read(65536)
+                chunk = hf_resp.read(1048576)
                 if not chunk:
                     break
                 if tmpf: