Cache.add() encountered a network error

[thekevinscott]

I'm seeing a Cache.add() error when trying to load Llama-3-8B-Instruct-q4f16_1-MLC:

Error: Cannot fetch https://huggingface.co/mlc-ai/Llama-3-8B-Instruct-q4f16_1-MLC/resolve/main/params_shard_0.bin err= NetworkError: Failed to execute 'add' on 'Cache': Cache.add() encountered a network error

This is Chrome 124.0.6367.119, running the demo at https://webllm.mlc.ai/

[thekevinscott]

It looks like any model I haven't previously cached is reporting the error:

[thekevinscott]

I don't think this is related to the typical browser cache, but just for good measure I cleared my cache, but same issue.

[ucalyptus2]

@thekevinscott I'm facing the same issue on chrome v124

i cleared my cache but no improvement

[ucalyptus2]

cc: @CharlieFRuan

[CharlieFRuan]

Is this issue encountered for all models? To be honest the Cache.add() is a bit vague; I've encountered this when the URL is wrong, but it does not seem to be the case here.

To triage a bit, could you check, in the console, application -> Cache storage, and see if webllm/config or webllm/wasm are populated? Just wanted to see if this is weight-downloading specific, or also applies to config json file and the wasm.

Besides, is this encountered for all models? Perhaps could you try, say TinyLlama? Wanted to know if it is due to a single weight shard being too large.