Upload logs/0895da42_tf_sftT5.log with huggingface_hub
Browse files
logs/0895da42_tf_sftT5.log
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
[transformers] This model config has set a `rope_parameters['original_max_position_embeddings']` field, to be used together with `max_position_embeddings` to determine a scaling factor. Please set the `factor` field of `rope_parameters`with this ratio instead -- we recommend the use of this field over `original_max_position_embeddings`, as it is compatible with most model architectures.
|
| 2 |
14:24:36 held-out rows 583 (of 600)
|
| 3 |
-
14:
|
| 4 |
-
14:
|
| 5 |
-
14:
|
|
|
|
| 1 |
[transformers] This model config has set a `rope_parameters['original_max_position_embeddings']` field, to be used together with `max_position_embeddings` to determine a scaling factor. Please set the `factor` field of `rope_parameters`with this ratio instead -- we recommend the use of this field over `original_max_position_embeddings`, as it is compatible with most model architectures.
|
| 2 |
14:24:36 held-out rows 583 (of 600)
|
| 3 |
+
14:35:47 TFM sft/sftT5_final.pt r=0 step 500: ans_gold 0.611 last_step 0.063 tok_acc 0.893 ans_nll 2.056 (n=583)
|
| 4 |
+
14:35:47 TFM_TASKS sft/sftT5_final.pt r=0 base_conversion:0.69/0.00 basic_arithmetic:0.94/0.00 calendar_arithmetic:0.64/0.05 chain_sum:1.00/0.07 count_primes:1.00/0.00 decimal_arithmetic:0.25/0.00 family_relationships:0.14/0.00 fraction_simplification:0.67/0.40 gcd:0.95/0.14 knights_knaves:0.30/0.13 lcm:0.96/0.04 leg_counting:1.00/0.00 letter_counting:0.85/0.04 number_filtering:0.05/0.00 number_sequence:0.89/0.05 number_sorting:0.00/0.00 prime_factorization:0.10/0.02 products:0.96/0.08 propositional_logic:0.56/0.22 simple_equations:1.00/0.00 spell_backward:0.00/0.00 syllogism:0.06/0.10 time_intervals:0.52/0.00 word_sorting:0.25/0.25
|
| 5 |
+
14:35:47 TFM DONE
|