Elizezen commited on
Commit
a0f145c
1 Parent(s): 0f1a3c8

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +15 -1
README.md CHANGED
@@ -24,6 +24,18 @@ In extensive testing and benchmarks, Omnia has proven to be an exceptionally str
24
  ## Benchmark Results
25
 
26
 
 
 
 
 
 
 
 
 
 
 
 
 
27
  **Benchmark Metrics:**
28
 
29
  ### Complexity
@@ -49,4 +61,6 @@ Similarity: 0.969
49
 
50
  ```
51
 
52
- *While the benchmark provides some insights, it is important to consider that the specific undisclosed details of the benchmark may introduce biases. Therefore, it is recommended to take this result with a grain of salt for now.*
 
 
 
24
  ## Benchmark Results
25
 
26
 
27
+ | Model | average | complexity | contextual maintenance | similarity to ground truth |
28
+ | -------------------------------------- | -------- | ---------- | ---------------------- | -------------------------- |
29
+ | **Omnia-2x7B** | **82.2** | **55.5** | **95.9** | **95.3** |
30
+ | Omnia-7B | 81.8 | 55.0 | 95.7 | 94.8 |
31
+ | Antler-RP-ja-westlake-chatvector | 81.6 | 54.2 | 95.6 | 95.0 |
32
+ | LightChatAssistant-TypeB-2x7B | 81.5 | 55.1 | 95.0 | 94.5 |
33
+ | Antler-7B | 81.5 | 53.7 | 95.8 | 95.1 |
34
+ | chatntq-ja-7b-v1.0-westlake-chatvector | 79.7 | 49.8 | 95.0 | 94.3 |
35
+ | chatntq-ja-7b-v1.0 | 79.7 | 52.0 | 93.9 | 93.2 |
36
+
37
+
38
+
39
  **Benchmark Metrics:**
40
 
41
  ### Complexity
 
61
 
62
  ```
63
 
64
+ *While the benchmark provides some insights, it is important to consider that the specific undisclosed details of the benchmark may introduce biases. Therefore, it is recommended to take this result with a grain of salt for now.*
65
+
66
+ If you want to know the full generation results of the benchmark, please contact me at [[email protected]](mailto:[email protected])