MTS AI Chat 7B
Created at 11.02.2024 22:10Assessment of the main tasks: 0.479
The table will scroll to the left
| Task name | Result | Metric |
|---|---|---|
| LCS | 0.094 | Accuracy |
| RCB | 0.532 / 0.53 | Avg. F1 / Accuracy |
| USE | 0.128 | Grade Norm |
| RWSD | 0.615 | Accuracy |
| PARus | 0.834 | Accuracy |
| ruTiE | 0.574 | Accuracy |
| MultiQ | 0.361 / 0.278 | F1-score/EM |
| CheGeKa | 0.083 / 0.046 | F1 / EM |
| ruModAr | 0.717 | EM |
| ruMultiAr | 0.233 | EM |
| MathLogicQA | 0.407 | Accuracy |
| ruWorldTree | 0.846 / 0.845 | Avg. F1 / Accuracy |
| ruOpenBookQA | 0.763 / 0.762 | Avg. F1 / Accuracy |
Evaluation on open tasks:
It is not taken into account in the overall ratingThe table will scroll to the left
| Task name | Result | Metric | ||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| BPS | 0.276 | Accuracy | ||||||||||||||||||||||||
| ruMMLU | 0.689 | Accuracy | ||||||||||||||||||||||||
| SimpleAr | 0.955 | EM | ||||||||||||||||||||||||
| ruHumanEval | 0.018 / 0.088 / 0.177 | pass@k | ||||||||||||||||||||||||
| ruHHH |
0.719
|
Accuracy | ||||||||||||||||||||||||
| ruHateSpeech |
0.758
|
Accuracy | ||||||||||||||||||||||||
| ruDetox |
|
Overall average score (J) Assessment of the preservation of meaning (SIM) Assessment of naturalness (FL) Style Transfer Accuracy (STA) |
||||||||||||||||||||||||
| ruEthics |
Table results:
[[-0.276, -0.28
, -0.279, -0.247
, -0.223], |
5 MCC |
Information about the submission:
Team:
MTS AI
Name of the ML model:
MTS AI Chat 7B
Link to the ML model:
https://huggingface.co/mistralai/Mistral-7B-v0.1Architecture description:
Mistral 7B model architecture
Description of the training:
Mistral trained on proprietary DPO and SFT datasets
Pretrain data:
-
Training Details:
-
License:
Proprietary model developed by MTS AI
Strategy, generation and parameters:
Code version v.1.1.0 All the parameters were not changed. Inference details: torch 2.1.0 + Cuda 11.8. max length 6012 tokens
Comments about inference:
we run the model using MERA github repo without any changes using hf inference script
