Allow readable profile names and configurable KV cache

test_auto_test.py

@@ -22,6 +22,15 @@ class AutoTestsTests(unittest.TestCase):
   self.assertIn((75,25),tested_splits);self.assertIn((50,50),tested_splits)
   self.auto.save(dict(job_id=job['id'],index=0,name='saved'));self.profiles.save.assert_called_once()
   self.assertEqual(AutoTests(self.auto.path,self.profiles,self.worker,self.scheduler).job['state'],'complete')
+ def test_kv_cache_choice_is_used_and_saved(self):
+  self.auto.request=Mock(return_value={});self.auto.benchmark=Mock(return_value={'predicted_per_second':40})
+  with patch('auto_test.probe',return_value=self.gpus):
+   self.auto.start(dict(model_id='m',start_context=4096,max_context=4096,mtp=False,cache_type_k='q8_0',cache_type_v='f16'))
+  self.auto.thread.join(3)
+  row=next(r for r in self.auto.job['results'] if r['success'])
+  self.assertEqual((row['parameters']['cache_type_k'],row['parameters']['cache_type_v']),('q8_0','f16'))
+  self.auto.save(dict(job_id=self.auto.job['id'],index=self.auto.job['results'].index(row),name='KV Test'))
+  self.assertEqual(self.profiles.save.call_args.args[0]['parameters']['cache_type_v'],'f16')
  def test_secondary_before_cpu(self):
   def plan(p,cancel):
    if p['parameters']['gpu_offload']=='full':raise InferenceError('does not fit')