bittensor
bittensor copied to clipboard
opentensor
Internet-scale Neural Networks
This PR adds a CUDA solver for the registration POW. - Requires: [cubit](https://github.com/opentensor/cubit/) - see requirements for cubit - installed as extra `pip install -e .[cubit]` - will throw error...
Turns out the bit shift to unpack the difficulty was also wrong - Fixed the bit shift - Added a test to catch this - Added a process restriction on...
#876 Fixes the flags The CPU register optimization introduced a bug where the btcli run command lacked some config flags that were used in CLI.register. This PR adds them in
- Moves overview args to cli parsers - Fixes overview when using `--hotkeys` or `--wallet.hotkeys`
- Fixed an issue where the SynapseType was TextCausalLMNext, when it should be TEXT_CAUSAL_LM_NEXT.
Utilizing the Accelerate package from HuggingFace allows us to run larger models than 6.7B.
**Is your feature request related to a problem? Please describe.** A clear and concise description of what the problem is. Ex. I'm always frustrated when [...] I'm looking for multi-sig...
Adds text generation as a function on the dendrite: # The below code prompts the best miner on the network. import bittensor wallet = bittensor.wallet() graph = bittensor.metagraph().sync() print (...
This PR addresses the issues with CUDA registration introduced by v3.3.4 This fixes: - Improper termination And this should fix: - Reduced actual hashrate (does not match reported) - `nonce_start...
This PR speeds up the storage call made by`subtensor.neurons` in the `subtensor.use_neurons_fast` function. This feature works by bundling a nodejs binary with the [polkadotjs API](https://github.com/polkadot-js/api). This binary is a CLI...