Update README.md
Browse files
README.md
CHANGED
@@ -20,7 +20,7 @@ widget:
|
|
20 |
example_title: "fashion - Chinese"
|
21 |
- text: "Wat zei je over mijn moeder? peter szemraj: Ik"
|
22 |
example_title: "🚎 - Dutch"
|
23 |
-
- text: "
|
24 |
example_title: "brain teaser - Polish"
|
25 |
- text: "Minha amiga diz que conhece todas as línguas, mas não fala nenhuma delas... o que há de errado com ela? peter szemraj: Eu"
|
26 |
example_title: "language - Portuguese"
|
@@ -43,7 +43,9 @@ widget:
|
|
43 |
- text: "как написать хорошие подсказки для языковых моделей? peter szemraj: ну, во-первых,"
|
44 |
example_title: "prompt tutorial - Russian"
|
45 |
- text: "Pewien mężczyzna wpycha swój samochód do hotelu i mówi właścicielowi, że jest bankrutem. Dlaczego? peter szemraj: może"
|
46 |
-
example_title: "brain teaser - Polish"
|
|
|
|
|
47 |
|
48 |
inference:
|
49 |
parameters:
|
@@ -68,6 +70,11 @@ inference:
|
|
68 |
|
69 |
- testing if fine-tuned personality data bleeds over to other languages without being trained in them explicitly
|
70 |
|
|
|
|
|
|
|
|
|
|
|
71 |
### Usage in python
|
72 |
|
73 |
Install the transformers library if you don't have it:
|
@@ -107,7 +114,7 @@ The following hyperparameters were used during training:
|
|
107 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
108 |
- lr_scheduler_type: cosine_with_restarts
|
109 |
- lr_scheduler_warmup_ratio: 0.05
|
110 |
-
- num_epochs:
|
111 |
|
112 |
### Framework versions
|
113 |
|
|
|
20 |
example_title: "fashion - Chinese"
|
21 |
- text: "Wat zei je over mijn moeder? peter szemraj: Ik"
|
22 |
example_title: "🚎 - Dutch"
|
23 |
+
- text: "Zagadka: Najpierw mnie zjadasz, a potem sam zostajesz zjedzony. Czym ja jestem? peter szemraj: Czy to"
|
24 |
example_title: "brain teaser - Polish"
|
25 |
- text: "Minha amiga diz que conhece todas as línguas, mas não fala nenhuma delas... o que há de errado com ela? peter szemraj: Eu"
|
26 |
example_title: "language - Portuguese"
|
|
|
43 |
- text: "как написать хорошие подсказки для языковых моделей? peter szemraj: ну, во-первых,"
|
44 |
example_title: "prompt tutorial - Russian"
|
45 |
- text: "Pewien mężczyzna wpycha swój samochód do hotelu i mówi właścicielowi, że jest bankrutem. Dlaczego? peter szemraj: może"
|
46 |
+
example_title: "brain teaser - Polish 2"
|
47 |
+
- text: "Zagadka: Mówię bez ust i słyszę bez uszu. Nie mam ciała, ale ożywiam się wraz z wiatrem. Czym jestem? peter szemraj: Czy to"
|
48 |
+
example_title: "brain teaser - Polish 3"
|
49 |
|
50 |
inference:
|
51 |
parameters:
|
|
|
70 |
|
71 |
- testing if fine-tuned personality data bleeds over to other languages without being trained in them explicitly
|
72 |
|
73 |
+
**Interesting findings thus far:**
|
74 |
+
|
75 |
+
- Passing a generic word after the `<name-identifier>` that is in a non-English language helps ensure the model responds in the question language (see: any example).
|
76 |
+
- Model generations (in general) remain semantically consistent, even if the generations switch from `<language>`to English in the middle of the generated text. This demonstrates some sort of "universal concept understanding" .
|
77 |
+
|
78 |
### Usage in python
|
79 |
|
80 |
Install the transformers library if you don't have it:
|
|
|
114 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
115 |
- lr_scheduler_type: cosine_with_restarts
|
116 |
- lr_scheduler_warmup_ratio: 0.05
|
117 |
+
- num_epochs: 1 (in addition to all training on prior checkpoints)
|
118 |
|
119 |
### Framework versions
|
120 |
|