t

app.py

@@ -105,12 +105,12 @@ def f(context, temperature, top_p, max_length, model_idx, SPACE_VERIFICATION_KEY
                 # for this api, a length of > 250 instantly errors, so use a while loop or something
                 # that would fetch results in chunks of 250
                 # NOTE change so it uses previous generated input every time
-                _context = context
+                # _context = context
                 generated_text = ""#context #""
                 while (max_length > 0):
                     # context becomes the previous generated context
                     # NOTE I've set return_full_text to false, see how this plays out
-                    payload = {"inputs": context, "parameters": {"return_full_text":False, "max_new_tokens": max_length>250 and 250 or max_length, "temperature": temperature, "top_p": top_p}}
+                    payload = {"inputs": context, "parameters": {"max_new_tokens": max_length>250 and 250 or max_length, "temperature": temperature, "top_p": top_p}}
                     response = requests.request("POST", API_URL, data=json.dumps(payload), headers=headers)
                     context = json.loads(response.content.decode("utf-8"))#[0]['generated_text']
                     # context = get_generated_text(generated_context)
@@ -131,7 +131,7 @@ def f(context, temperature, top_p, max_length, model_idx, SPACE_VERIFICATION_KEY
                 # data = json.dumps(payload)
                 # response = requests.request("POST", API_URL, data=data, headers=headers)
                 # generated_text = json.loads(response.content.decode("utf-8"))[0]['generated_text']
-                return _context+generated_text
+                return generated_text#context #_context+generated_text
 
             # use secondary gpt-j-6B api, as the main one is down
             if not secondary_gpt_j_api_up: