diff --git a/litellm/main.py b/litellm/main.py
index df2b3ce2382..3f236523995 100644
--- a/litellm/main.py
+++ b/litellm/main.py
@@ -1792,11 +1792,21 @@ def text_completion(*args, **kwargs):
         if kwargs["model"] in litellm.open_ai_text_completion_models and response._hidden_params.get("original_response", None) != None:
             return response._hidden_params.get("original_response", None)
         transformed_logprobs = None
+        # only supported for TGI models
         try:
             raw_response = response._hidden_params.get("original_response", None)
+            tokens = []
+            token_logprobs = []
+            if "prefill" in raw_response[0]["details"]:
+                prefills = raw_response[0]["details"]['prefill']
+                for prefill in prefills:
+                    tokens.append(prefill['text'])
+                    token_logprobs.append(prefill['logprob'])
+            new_tokens = [token['text'] for token in raw_response[0]['details']['tokens']]
+            new_token_logprobs = [token['logprob'] for token in raw_response[0]['details']['tokens']]         
             transformed_logprobs = {
-                "tokens": [token['text'] for token in raw_response[0]['details']['tokens']],
-                "token_logprobs": [token['logprob'] for token in raw_response[0]['details']['tokens']]
+                "tokens": tokens + new_tokens,
+                "token_logprobs": token_logprobs + new_token_logprobs
             }
         except Exception as e:
             print("LiteLLM non blocking exception", e)
diff --git a/litellm/utils.py b/litellm/utils.py
index d7aa6a0a8f9..1f2145e52ef 100644
--- a/litellm/utils.py
+++ b/litellm/utils.py
@@ -1348,6 +1348,10 @@ def get_optional_params(  # use the openai defaults
             optional_params["best_of"] = n
         if presence_penalty:
             optional_params["repetition_penalty"] = presence_penalty
+        if "echo" in special_params:
+            # https://huggingface.co/docs/huggingface_hub/main/en/package_reference/inference_client#huggingface_hub.InferenceClient.text_generation.decoder_input_details
+            #  Return the decoder input token logprobs and ids. You must set details=True as well for it to be taken into account. Defaults to False
+            optional_params["decoder_input_details"] = special_params["echo"]
     elif custom_llm_provider == "together_ai":
         ## check if unsupported param passed in 
         supported_params = ["stream", "temperature", "max_tokens", "top_p", "stop", "frequency_penalty"]