Skip to content

Conversation

@rishsriv
Copy link
Member

We add a deepseek runner to support evaluating deepseek models via their official API.

To verify that this works

deepseek-chat

python main.py \      
  -db postgres \
  -q "data/questions_gen_postgres.csv" "data/instruct_basic_postgres.csv" "data/instruct_advanced_postgres.csv" \
  -o results/deepseek_classic.csv results/deepseek_basic.csv results/deepseek_advanced.csv \
  -g deepseek \
  -f prompts/prompt_openai.json \
  -m deepseek-chat \
  -p 5 \
  -c 0

For v1

Correct so far: 188/210 (89.52%)
                 correct  error_db_exec
query_category                         
date_functions  0.800000       0.028571
group_by        1.000000       0.000000
instruct        0.914286       0.000000
order_by        0.914286       0.000000
ratio           0.771429       0.057143
table_join      0.971429       0.000000

For basic

Correct so far: 38/40 (95.00%)
                                   correct  error_db_exec
query_category                                           
basic_group_order_limit              1.000            0.0
basic_join_date_group_order_limit    0.875            0.0
basic_join_distinct                  1.000            0.0
basic_join_group_order_limit         0.875            0.0
basic_left_join                      1.000            0.0

For advanced

Correct so far: 54/64 (84.38%)
                              correct  error_db_exec
query_category                                      
instructions_cte_join          0.9375          0.000
instructions_cte_window        0.6250          0.000
instructions_date_join         0.7500          0.125
instructions_string_matching   1.0000          0.000
keywords_aggregate             0.8750          0.000
keywords_ratio                 0.8750          0.125

deepseek-reasoner

python main.py \      
  -db postgres \
  -q "data/questions_gen_postgres.csv" "data/instruct_basic_postgres.csv" "data/instruct_advanced_postgres.csv" \
  -o results/deepseek_classic.csv results/deepseek_basic.csv results/deepseek_advanced.csv \
  -g deepseek \
  -f prompts/prompt_openai_o1.json \
  -m deepseek-reasoner \
  -p 5 \
  -c 0

Taking too long to finish executing, but runs!

@rishsriv rishsriv merged commit fa1c76f into main Jan 21, 2025
2 checks passed
@rishsriv rishsriv deleted the rishabh/deepseek-support branch January 21, 2025 08:34
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants