[tfjs-converter] Allow custom shard size for more conversion pairs

[pvaneck]

Previously, only the tfjs_layers_model to tfjs_layers_model conversion pair worked with the weight_shard_size_bytes flag. This PR allows the flag to be used for any conversion where the output_format is tfjs_layers_model or tfjs_graph_model.

---

This change is 

[pyu10055]

Thank you for the great contribution. The PR looks great, one high level comment, we have the wizard.ts file https://github.com/tensorflow/tfjs/blob/master/tfjs-converter/python/tensorflowjs/converters/wizard.py
can you add support for the shard size param too? thanks

Reviewed 1 of 6 files at r1.
Reviewable status: 0 of 1 approvals obtained (waiting on @pyu10055)

[pvaneck]

@pyu10055 Thanks for the review. I added support for the shard size param in the wizard.

[pyu10055]

Reviewable status: 0 of 1 approvals obtained (waiting on @pvaneck)

---

tfjs-converter/python/tensorflowjs/converters/wizard.py, line 486 at r2 (raw file):

          'when': lambda answers: (value_in_list(answers, common.OUTPUT_FORMAT,
                                                 (common.TFJS_LAYERS_MODEL,
                                                  common.TFJS_GRAPH_MODEL)) or

should this be 'and' instead of 'or'?

[pvaneck]

Reviewable status: 0 of 1 approvals obtained (waiting on @pyu10055)

---

tfjs-converter/python/tensorflowjs/converters/wizard.py, line 486 at r2 (raw file):

Previously, pyu10055 (Ping Yu) wrote…

should this be 'and' instead of 'or'?

Ideally, we would only need the first condition before the or, however, if the input format is TF_SAVED_MODEL or TF_HUB_MODEL, the implication is that the model will be converted to TFJS_GRAPH_MODEL. Since this is implied, the user isn't asked in the wizard to select an output format, so the OUTPUT_FORMAT key is never set in answers. Thus, the first condition where we just check OUTPUT_FORMAT would fail.

So, that second or condition is just another way of checking if the OUTPUT_FORMAT is TFJS_GRAPH_MODEL based on the input format.

[pyu10055]

Reviewable status: complete! 1 of 1 approvals obtained (waiting on @pvaneck)

---

tfjs-converter/python/tensorflowjs/converters/wizard.py, line 486 at r2 (raw file):

Previously, pvaneck (Paul Van Eck) wrote…

Ideally, we would only need the first condition before the or, however, if the input format is TF_SAVED_MODEL or TF_HUB_MODEL, the implication is that the model will be converted to TFJS_GRAPH_MODEL. Since this is implied, the user isn't asked in the wizard to select an output format, so the OUTPUT_FORMAT key is never set in answers. Thus, the first condition where we just check OUTPUT_FORMAT would fail.

So, that second or condition is just another way of checking if the OUTPUT_FORMAT is TFJS_GRAPH_MODEL based on the input format.

make sense, thank you.